INTRODUCTION TO THE SERIES
The aim of the Handbooks in Economics series is to produce Handbooks for various branches of economics, each of which is a definitive source, reference, and teaching supplement for use by professional researchers and advanced graduate students. Each Handbook provides self-contained surveys of the current state of a branch of economics in the form of chapters prepared by leading specialists on various aspects of this branch of economics. These surveys summarize not only received results but also newer developments, from recent journal articles and discussion papers. Some original material is also included, but the main goal is to provide comprehensive and accessible surveys. The Handbooks are intended to provide not only useful reference volumes for professional collections but also possible supplementary readings for advanced courses for graduate students in economics. KENNETH J. ARROW and MICHAEL D. INTRILIGATOR
PUBLISHER’S NOTE For a complete overview of the Handbooks in Economics Series, please refer to the listing at the end of this volume.
v
CONTENTS OF VOLUME 1A
Introduction to the Series
v
Contents of the Handbook
vii
Preface to the Handbook of Economic Growth
xi
INTRODUCTION: GROWTH IN RETROSPECT AND PROSPECT Reflections on Growth Theory ROBERT M. SOLOW Abstract Keywords
3 3 3 3
PART I: THEORIES OF ECONOMIC GROWTH Chapter 1 Neoclassical Models of Endogenous Growth: The Effects of Fiscal Policy, Innovation and Fluctuations LARRY E. JONES AND RODOLFO E. MANUELLI Abstract Keywords 1. Introduction 2. Endogenous growth: Infinite lifetimes 2.1. Growth and the Solow model 2.2. A one sector model of equilibrium growth 2.3. Fiscal policy and growth 2.4. Innovation in the neoclassical model
3. Fluctuations and growth 3.1. Introduction 3.2. Empirical evidence 3.3. Theoretical models 3.4. A simple linear endogenous growth model 3.5. Physical and human capital 3.6. The opportunity cost view 3.7. More on government spending, taxation, and growth 3.8. Quantitative effects
4. Concluding comments xvii
13 14 14 15 16 17 19 21 27 32 32 32 39 39 49 54 56 59 62
xviii
Contents of Volume 1A
Acknowledgements References Chapter 2 Growth with Quality-Improving Innovations: An Integrated Framework PHILIPPE AGHION AND PETER HOWITT Abstract Keywords 1. Introduction 2. Basic framework 2.1. 2.2. 2.3. 2.4.
A toy version of the Aghion–Howitt model A generalization Alternative formulations Comparative statics on growth
3. Linking growth to development: convergence clubs 3.1. 3.2. 3.3. 3.4.
A model of technology transfer World growth and distribution The role of financial development in convergence Concluding remark
4. Linking growth to IO: innovate to escape competition 4.1. The theory 4.2. Empirical predictions 4.3. Empirical evidence and relationship to literature 4.4. A remark on inequality and growth
5. Scale effects 5.1. Theory 5.2. Evidence 5.3. Concluding remarks
6. Linking growth to institutional change 6.1. From Schumpeter to Gerschenkron 6.2. A simple model of appropriate institutions 6.3. Appropriate education systems
7. Conclusion References Chapter 3 Horizontal Innovation in the Theory of Growth and Development GINO GANCIA AND FABRIZIO ZILIBOTTI Abstract Keywords 1. Introduction 2. Growth with expanding variety 2.1. The benchmark model
63 63
67 68 68 69 69 69 71 75 75 76 78 81 81 84 84 86 89 89 92 92 92 94 97 98 98 100 101 106 107
111 112 112 113 116 116
Contents of Volume 1A 2.2. Two variations of the benchmark model: “lab-equipment” and “labor-for intermediates” 2.3. Limited patent protection
3. Trade, growth and imitation 3.1. Scale effects, economic integration and trade 3.2. Innovation, imitation and product cycles
4. Directed technical change 4.1. Factor-biased innovation and wage inequality 4.2. Appropriate technology and development 4.3. Trade, inequality and appropriate technology
5. Complementarity in innovation 6. Financial development 7. Endogenous fluctuations 7.1. Deterministic cycles 7.2. Learning and sunspots
8. Conclusions Acknowledgements References Chapter 4 From Stagnation to Growth: Unified Growth Theory ODED GALOR Abstract Keywords 1. Introduction 2. Historical evidence 2.1. The Malthusian epoch 2.2. The Post-Malthusian Regime 2.3. The Sustained Growth Regime 2.4. The great divergence
3. The fundamental challenges 3.1. Mysteries of the growth process 3.2. The incompatibility of non-unified growth theories 3.3. Theories of the demographic transition and their empirical assessment
4. Unified growth theory 4.1. From stagnation to growth 4.2. Complementary theories
5. Unified evolutionary growth theory 5.1. Human evolution and economic development 5.2. Natural selection and the origin of economic growth 5.3. Complementary mechanisms
6. Differential takeoffs and the great divergence 6.1. Non-unified theories 6.2. Unified theories
xix
120 122 124 124 127 130 131 136 140 144 150 157 158 162 166 166 166
171 172 173 174 178 179 185 195 218 219 220 221 224 235 237 256 264 264 266 273 276 277 279
xx
Contents of Volume 1A
7. Concluding remarks Acknowledgements References Chapter 5 Poverty Traps COSTAS AZARIADIS AND JOHN STACHURSKI Abstract Keywords 1. Introduction 2. Development facts 2.1. Poverty and riches 2.2. A brief history of economic development
3. Models and definitions 3.1. Neoclassical growth with diminishing returns 3.2. Convex neoclassical growth and the data 3.3. Poverty traps: historical self-reinforcement 3.4. Poverty traps: inertial self-reinforcement
4. Empirics of poverty traps 4.1. Bimodality and convergence clubs 4.2. Testing for existence 4.3. Model calibration 4.4. Microeconomic data
5. Nonconvexities, complementarities and imperfect competition 5.1. Increasing returns and imperfect competition 5.2. The financial sector and coordination 5.3. Matching 5.4. Other studies of increasing returns
6. Credit markets, insurance and risk 6.1. Credit markets and human capital 6.2. Risk 6.3. Credit constraints and endogenous inequality
7. Institutions and organizations 7.1. Corruption and rent-seeking 7.2. Kinship systems
8. Other mechanisms 9. Conclusions 9.1. Lessons for economic policy
Acknowledgements Appendix A: A.1. Markov chains and ergodicity A.2. Remaining proofs
References
283 285 285
295 296 296 297 303 303 304 307 307 312 317 326 330 330 335 337 339 340 341 343 346 349 350 351 355 358 363 364 367 373 373 374 375 375 375 378 379
Contents of Volume 1A
Chapter 6 Institutions as a Fundamental Cause of Long-Run Growth DARON ACEMOGLU, SIMON JOHNSON AND JAMES A. ROBINSON Abstract Keywords 1. Introduction 1.1. The question 1.2. The argument 1.3. Outline
2. Fundamental causes of income differences 2.1. Three fundamental causes
3. Institutions matter 3.1. The Korean experiment 3.2. The colonial experiment
4. The Reversal of Fortune 4.1. 4.2. 4.3. 4.4. 4.5. 4.6.
The reversal among the former colonies Timing of the reversal Interpreting the reversal Economic institutions and the reversal Understanding the colonial experience Settlements, mortality and development
5. Why do institutions differ? 5.1. 5.2. 5.3. 5.4.
The efficient institutions view – the Political Coase Theorem The ideology view The incidental institutions view The social conflict view
6. Sources of inefficiencies 6.1. 6.2. 6.3. 6.4. 6.5.
Hold-up Political losers Economic losers The inseparability of efficiency and distribution Comparative statics
6.6. The colonial experience in light of the comparative statics 6.7. Reassessment of the social conflict view
7. The social conflict view in action 7.1. Labor markets 7.2. Financial markets 7.3. Regulation of prices 7.4. Political power and economic institutions
8. A theory of institutions 8.1. Sources of political power 8.2. Political power and political institutions 8.3. A theory of political institutions
xxi
385 386 387 388 388 389 396 396 397 402 404 407 407 408 412 412 414 416 417 421 422 424 425 427 428 430 432 434 436 437 438 439 439 440 441 443 445 448 448 449 451
xxii
Contents of Volume 1A
9. The theory in action 9.1. Rise of constitutional monarchy and economic growth in early modern Europe 9.2. Summary 9.3. Rise of electoral democracy in Britain 9.4. Summary
10. Future avenues Acknowledgements References Chapter 7 Growth Theory through the Lens of Development Economics ABHIJIT V. BANERJEE AND ESTHER DUFLO Abstract Keywords 1. Introduction: neo-classical growth theory 1.1. The aggregate production function 1.2. The logic of convergence
2. Rates of return and investment rates in poor countries 2.1. Are returns higher in poor countries? 2.2. Investment rates in poor countries
3. Understanding rates of return and investment rates in poor countries: aggregative approaches 3.1. 3.2. 3.3. 3.4.
Access to technology and the productivity gap Human capital externalities Coordination failure Taking stock
4. Understanding rates of return and investment rates in poor countries: nonaggregative approaches 4.1. 4.2. 4.3. 4.4.
Government failure The role of credit constraints Problems in the insurance markets Local externalities
4.5. The family: incomplete contracts within and across generations 4.6. Behavioral issues
5. Calibrating the impact of the misallocation of capital 5.1. A model with diminishing returns 5.2. A model with fixed costs
6. Towards a non-aggregative growth theory 6.1. An illustration 6.2. Can we take this model to the data? 6.3. Where do we go from here?
Acknowledgements References
452 452 457 458 462 463 464 464
473 474 474 475 475 477 479 479 493 499 499 501 503 504 505 505 509 512 515 518 520 522 523 527 535 535 538 542 544 544
Contents of Volume 1A
xxiii
PART II: EMPIRICS OF ECONOMIC GROWTH Chapter 8 Growth Econometrics STEVEN N. DURLAUF, PAUL A. JOHNSON AND JONATHAN R.W. TEMPLE Abstract Keywords 1. Introduction 2. Stylized facts 2.1. 2.2. 2.3. 2.4. 2.5. 2.6. 2.7. 2.8. 2.9. 2.10.
A long-run view Data after 1960 Differences in levels of GDP per worker Growth miracles and disasters Convergence? The growth slowdown Does past growth predict future growth? Growth differences by development level and geographic region Stagnation and output volatility A summary of the stylized facts
3. Cross-country growth regressions: from theory to empirics 3.1. Growth dynamics: basic ideas 3.2. Cross-country growth regressions 3.3. Interpreting errors in growth regressions
4. The convergence hypothesis 4.1. Convergence and initial conditions 4.2. β-convergence 4.3. Distributional approaches to convergence 4.4. Time series approaches to convergence 4.5. Sources of convergence or divergence
5. Statistical models of the growth process 5.1. Specifying explanatory variables in growth regressions 5.2. Parameter heterogeneity 5.3. Nonlinearity and multiple regimes
6. Econometric issues I: Alternative data structures 6.1. 6.2. 6.3. 6.4.
Time series approaches Panel data Event study approaches Endogeneity and instrumental variables
7. Econometric issues II: Data and error properties 7.1. 7.2. 7.3. 7.4.
Outliers Measurement error Missing data Heteroskedasticity
555 556 557 558 561 562 562 563 565 567 567 568 571 573 575 576 576 578 581 582 582 585 592 599 604 607 608 616 619 624 624 627 636 637 640 641 641 642 643
xxiv
Contents of Volume 1A
7.5. Cross-section error correlation
8. Conclusions: The future of growth econometrics Acknowledgements Appendix A: Data Key to the 102 countries Extrapolation
Appendix B: Variables in cross-country growth regressions Appendix C: Instrumental variables for Solow growth determinants Appendix D: Instrumental variables for non-Solow growth determinants References Chapter 9 Accounting for Cross-Country Income Differences FRANCESCO CASELLI Abstract 1. Introduction 2. The measure of our ignorance 2.1. 2.2. 2.3. 2.4.
Basic data Basic measures of success Alternative measures used in the literature Sub-samples
3. Robustness: basic stuff 3.1. 3.2. 3.3. 3.4.
Depreciation rate Initial capital stock Education-wage profile Years of education 1
3.5. Years of education 2 3.6. Hours worked 3.7. Capital share
4. Quality of human capital 4.1. Quality of schooling: Inputs 4.2. Quality of schooling: test scores 4.3. Experience 4.4. Health 4.5. Social vs. private returns to schooling and health
5. Quality of physical capital 5.1. Composition 5.2. Vintage effects 5.3. Further problems with K
6. Sectorial differences in TFP 6.1. Industry studies 6.2. The role of agriculture 6.3. Sectorial composition and development accounting
643 645 651 651 651 652 652 660 661 663
679 680 681 683 685 686 688 689 690 690 691 693 694 694 695 696 698 698 703 706 708 710 711 711 715 716 717 718 719 724
Contents of Volume 1A
7. Non-neutral differences in technology 7.1. Basic concepts and qualitative results 7.2. Development accounting with non-neutral differences
8. Conclusions Acknowledgements References Chapter 10 Accounting for Growth in the Information Age DALE W. JORGENSON Abstract Keywords 1. The information age 1.1. Introduction 1.2. Faster, better, cheaper 1.3. Impact of information technology
2. Aggregate growth accounting 2.1. The role of information technology 2.2. The American growth resurgence 2.3. Demise of traditional growth accounting
3. International comparisons 3.1. 3.2. 3.3. 3.4. 3.5.
Introduction Investment and total factor productivity Investment in information technology Alternative approaches Conclusions
4. Economics on internet time Acknowledgements References Further reading Chapter 11 Externalities and Growth PETER J. KLENOW AND ANDRÉS RODRÍGUEZ-CLARE Abstract Keywords 1. Introduction 2. A brief guide to externalities in growth models 2.1. 2.2. 2.3. 2.4.
Models with knowledge externalities Models with knowledge externalities and new-good externalities Models with new-good externalities Models with no externalities
3. Cross-country evidence
xxv
727 727 734 737 738 738
743 744 745 746 746 747 755 756 756 765 779 784 784 787 796 803 805 806 807 808 815
817 818 818 819 819 820 821 822 823 825
xxvi 3.1. 3.2. 3.3. 3.4.
Contents of Volume 1A The world-wide growth slowdown Beta convergence in the OECD Low persistence of growth rate differences Investment rates and growth vs. levels
3.5. R&D and TFP
4. Models with common growth driven by international knowledge spillovers 4.1. R&D investment and relative productivity 4.2. Modeling growth in the world technology frontier 4.3. Determinants of R&D investment 4.4. Calibration 4.5. The benefits of engagement 4.6. Discussion of main results
5. Conclusion Acknowledgements Appendix A: Comparative statics
References
825 827 829 831 833 835 836 839 843 845 854 856 856 857 857 858 859
PART III: GROWTH POLICIES AND MECHANISMS Chapter 12 Finance and Growth: Theory and Evidence ROSS LEVINE Abstract Keywords 1. Introduction 2. Financial development and economic growth: Theory 2.1. 2.2. 2.3. 2.4. 2.5.
What is financial development? Producing information and allocating capital Monitoring firms and exerting corporate governance Risk amelioration Pooling of savings
2.6. Easing exchange 2.7. The theoretical case for a bank-based system 2.8. The theoretical case for a market-based system 2.9. Countervailing views to bank-based vs. market-based debate 2.10.Finance, income distribution, and poverty
3. Evidence on finance and growth 3.1. 3.2. 3.3. 3.4. 3.5.
Cross-country studies of finance and growth Panel, time-series, and case-studies of finance and growth Industry and firm level studies of finance and growth Are bank- or market-based systems better? Evidence Finance, income distribution, and poverty alleviation: evidence
865 866 866 867 869 869 870 872 875 879 880 881 883 886 887 888 889 899 910 918 920
Contents of Volume 1A
xxvii
4. Conclusions Acknowledgements References
921 923 923
Chapter 13 Human Capital and Technology Diffusion JESS BENHABIB AND MARK M. SPIEGEL Abstract Keywords 1. Introduction 2. Variations on the Nelson–Phelps model 3. Some microfoundations based on the diffusion model of Barro and Sala-iMartin 4. A nested specification 5. Empirical evidence 5.1. Measurement of total factor productivity 5.2. Model specification 5.3. Results
6. Model prediction 6.1. Model forecasting 6.2. Negative catch-up countries
7. Conclusion References Chapter 14 Growth Strategies DANI RODRIK Abstract Keywords 1. Introduction 2. What we know that (possibly) ain’t so 3. The indeterminate mapping from economic principles to institutional arrangements 4. Back to the real world 4.1. In practice, growth spurts are associated with a narrow range of policy reforms 4.2. The policy reforms that are associated with these growth transitions typically combine elements of orthodoxy with unorthodox institutional practices 4.3. Institutional innovations do not travel well 4.4. Sustaining growth is more difficult than igniting it, and requires more extensive institutional reform
5. A two-pronged growth strategy 5.1. An investment strategy to kick-start growth 5.2. An institution building strategy to sustain growth
935 936 936 937 940 944 946 948 948 953 954 959 959 960 964 965
967 968 968 969 973 978 989 989 993 994 996 997 998 1005
xxviii
Contents of Volume 1A
6. Concluding remarks Acknowledgements References
1009 1010 1010
Chapter 15 National Policies and Economic Growth: A Reappraisal WILLIAM EASTERLY Abstract Keywords 1. Theoretical models that predict strong policy effects 2. Models that predict small policy effects on growth 3. Empirics 4. Some empirical caveats 5. New empirical work 6. Policy episodes and transition paths 7. Institutions versus policies 8. Conclusions References
1015 1016 1016 1017 1026 1032 1033 1036 1050 1054 1056 1056
Author Index
I-1
Subject Index
I-37
PREFACE TO THE HANDBOOK OF ECONOMIC GROWTH
The progress which is to be expected in the physical sciences and arts, combined with the greater security of property, and greater security in disposing of it, which are obvious features in the civilization of modern nations, and with the more extensive and skillful employment of the joint-stock principle, afford space and scope for an indefinite increase of capital and production, and for the increase of population that is its ordinary accompaniment. John Stuart Mill, Principles of Political Economy, 1848 Interest is economic growth has been an integral part of economics since its inception as a scholarly discipline. Remarkably, this ancient lineage is consistent with growth economics representing one of the most active areas of research in economics in the last two decades. Perhaps more surprising, this activity followed a relatively long period of calm in the aftermath of the seminal theoretical and empirical work by Robert Solow on the neoclassical growth model [Solow (1956, 1957)]. Solow’s research set the growth research agenda for over 25 years. In terms of economic theory, much of the work of the 1960’s consisted of translating the Solow framework into an explicit intertemporal optimizing framework; this translation, enshrined in economics as the Cass–Koopmans model [David Cass (1965), Tjalling Koopmans (1965)] has been of great importance in much of the new growth theory as well. In terms of empirical work, Solow’s accounting framework stimulated many studies [a style of work well summarized in Edward Denison (1974)] which attempted more elaborate decompositions of growth patterns into components due to human and physical capital accumulation and a technology residual. Indeed, from the perspective of 1980, growth economics might have itself appeared to have achieved a steady state. This apparent steady state was shattered on both the theoretical and empirical levels in the late 1980’s and the 1990’s. In terms of theory, new models of endogenous growth1 questioned the neoclassical emphasis on capital accumulation as the main engine of growth, focusing instead on the Schumpeterian idea that growth is primarily driven by innovations that are themselves the result of profit-motivated research activities and create a conflict between the old and the new by making old technologies become obsolete. On the empirical side, Robert Barro (1991) and N. Gregory Mankiw,
1 Also based on capital accumulation are the so-called AK models of endogenous growth [Frankel (1962), Romer (1986), Lucas (1988)], in which capital accumulation generates knowledge accumulation. See the books by Grossman and Helpman (1991), Jones (2002), Barro and Sala-i-Martin (2003) or Aghion and Howitt (1998), for other references.
xi
xii
Preface to the Handbook of Economic Growth
David Romer, and David Weil (1992) launched the use of cross-country growth regressions to explore growth differences across countries; a cross-section that is far more extensive and covers much more of the world than occurred in earlier growth studies. These two parallel developments themselves gave birth to a whole range of new theoretical and empirical explorations of the determinants of growth and convergence – in particular the economic organizations and policies and the political institutions that are growth-enhancing at different stages of development. At the same time, new empirical methods were developed to reexamine issues of growth accounting on one end and which have begun to employ sophisticated statistical methods to uncover heterogeneities and nonlinearities on the other. This renaissance of growth economics reflects several factors. On the theory side, much of the work has been stimulated by modeling techniques imported in the 1970s from the new theory of international trade2 or the new theory of industrial organization,3 which made it possible to introduce imperfect competition and innovations in simple general equilibrium settings. Empirical work has been facilitated by the construction of new data sets, of which Alan Heston and Robert Summers [see Heston, Summers and Aten (2002) for the latest incarnation] has been especially influential. More recent work has made increasing use of new micro data, whether cross-industry, or cross-firm, or plant level. The availability of these new data sets, in turn has initiated a new phase in growth economics in which theory and empirics go hand in hand as the development of new growth theories generates or is itself prompted by the introduction of new statistical tools and empirical exercises. This phase is particularly exciting as one can more directly analyze the impact of specific institutional reforms or macroeconomic policies on economic growth across different types of countries. The Handbook of Economic Growth is designed to communicate the state of modern growth research. However, in contrast to other handbook volumes, we looked for chapters by active growth researchers. We then asked these authors to primarily convey the frontier ideas they are currently working on, anticipating that in order to put the reader up to speed with their current research agendas, the authors would also have to provide introductory surveys of contributions in their fields. As our readers will see, some chapters contain overlaps with other chapters and in a number of cases they partly disagree with one another. This only shows that growth economics is a lively field, with professional disagreements, alternative perspectives and outstanding controversies, but at the same time there exists a common eagerness to better understand the mechanics of economic development. The Handbook consists of 28 chapters and is divided into six parts. Part I lays out the theoretical foundations. The first chapter surveys the neo-classical and AK models of growth. The second chapter develops the Schumpeterian growth 2 See the product variety models of Romer (1990) and Grossman and Helpman (1990) and the whole litera-
ture that builds upon this approach, surveyed in Chapter 3 below. 3 The Schumpeterian models with quality-improving innovations, starting with Segerstrom, Anant and Dinopoulos (1990) and Aghion and Howitt (1992), belong to this second category.
Preface to the Handbook of Economic Growth
xiii
model with quality-improving innovations and confronts it with new empirical evidence. The third chapter surveys the literature that built upon Paul Romer’s productvariety model. The fourth chapter looks at growth in the very long run and analyzes the interplay between technical change and demographic transitions, and explores the issue of transitions between different growth regimes. The next chapter analyzes the central role of economic and political institutions, and describes the mechanisms whereby the dynamics of political institutions interacts with the dynamics of economic institutions and that of income inequality. The following chapter focuses on the emergence and existence of poverty traps, a question of particular importance in development contexts. The final chapter further explores the interplay of growth economics and development economics, with particular attention to how factors such as credit market constraints and intersectoral heterogeneity can explain outstanding puzzles concerning capital flows and interest rates, which are major elements of the growth process. Part II examines the empirics of growth. An important aspect of these chapters is the diversity of approaches that have been taken to link growth theory to data. Growth accounting continues to play an important role in growth economics, both in terms of organizing facts and in terms of identifying the domain in which new growth theories can supplement neoclassical explanations. Growth economics has at the same time stimulated the development of new econometrics tools to address the specific data implications of various growth theories, implications in some cases challenge the assumptions that underlie conventional econometric tools. One theme of the work in this Part of the Handbook is that there exist limits to what may be learned about the structural elements of the growth process from formal statistical models. At the same time, empirical growth work plays a key role in identifying the stylized facts that growth theories need to address. Part III of the Handbook examines a range of growth mechanisms. Some of these mechanisms have to do with the microeconomics of technology and education. Other mechanisms lie outside the domain of the neoclassical model and have to do with issues of political and economic institutions and social structure. Another theme that is developed here concerns the links between inequality and growth, which naturally raises issues of equity/efficiency tradeoffs. Finally, the role of government policy in affecting long run growth is studied. Much of the exciting work on growth has consisted of efforts to understand how factors beyond capital accumulation and technological change can affect growth; this very broad conception of the growth process is reflected in this section. Part IV explores a range of aspects concerning technology. The discussion starts with a chapter that reviews the history of technology from a growth perspective. This discussion is a valuable complement to the formal statistical analyses studied in Part II. The analysis then turns to alternative theories by which technology evolves and diffuses in an economy. General purpose technologies are studied as an engine of growth. The consequences of technological diffusion for economic transformations are described and the inequality consequences of technological change are considered. Finally, the role of technology barriers in producing persistent international inequality is examined.
xiv
Preface to the Handbook of Economic Growth
Part V considers the relationship between trade and geography. The discussion explores how trade and geographic agglomeration can affect growth trajectories as well as how growth interacts with geography to produce national boundaries. Further, some of the consequences of economic growth for a range of macroeconomic phenomena are explored in Part VI. Different chapters explore how growth affects inequality, sociological outcomes, and the environment. Finally, we are honored that Robert Solow has contributed a set of reflections on the state of growth economics to complete the Handbook. While growth economics has made immense strides in the last two decades, it is of course the case that the field “stands on the shoulders of giants”. And in this regard, Solow’s contributions are not alone. One can see the intertemporal optimization methodology that underlies the current theoretical analyses in the work of Frank Ramsey [Ramsey (1928)] and the ideas of social increasing returns in an early paper by Kenneth Arrow (1962). Such observations do not diminish the new growth economics, but rather speak well to the nature of progress in economics. We would like to thank Kenneth Arrow and Michael Intriligator for their support in initiating this project as well as in providing invaluable guidance throughout the process. Valerie Teng of North-Holland, Lauren LaRosa at Harvard and Alisenne Sumwalt at Wisconsin have provided terrific administrative assistance at various stages of this project. And of course, we are deeply grateful to the authors for their work. If nothing else, their contributions reinforce our view that the human capital contribution to production takes pride of place, at least when the growth of knowledge is concerned. Philippe Aghion and Steven Durlauf
References Aghion, P., Howitt, P. (1992). “A model of growth through creative destruction”. Econometrica 60, 323–351. Aghion, P., Howitt, P. (1998). Endogenous Growth Theory. MIT Press. Arrow, K. (1962). “The economic implications of learning by doing”. Review of Economic Studies 29, 155– 173. Barro, R. (1991). “Economic growth in a cross section of countries”. Quarterly Journal of Economics 106 (2), 407–443. Barro, R., Sala-i-Martin, X. (2003). Economic Growth. MIT Press. Cass, D. (1965). “Optimum growth in an aggregative model of capital accumulation”. Review of Economic Studies 32 (3), 233–240. Denison, E. (1974). Accounting for United States Economic Growth, 1929–1969. Brookings Institution Press, Washington, DC. Frankel, M. (1962). “The production function in allocation and growth: A synthesis”. American Economic Review 52, 995–1022. Grossman, G., Helpman, E. (1990). “Comparative advantage and long-run growth”. American Economic Review 80, 796–815. Grossman, G., Helpman, E. (1991). Innovations and Growth in the Global Economy. MIT Press. Heston, A., Summers, R., Aten, B. (2002). “Penn World Table Version 6.1”. Center for International Comparisons at the University of Pennsylvania (CICUP).
Preface to the Handbook of Economic Growth
xv
Jones, C. (2002). Introduction to Economic Growth, second ed. W.W. Norton, New York. Koopmans, T. (1965). “On the concept of optimal growth”. In: The Econometric Approach to Development Planning. Rand-McNally, Chicago. Lucas, R. (1988). “On the mechanics of economic development”. Journal of Monetary Economics 22, 3–42. Mankiw, N.G., Romer, D., Weil, D. (1992). “A contribution to the empirics of economic growth”. Quarterly Journal of Economics 107 (2), 407–437. Ramsey, F. (1928). “A mathematical theory of savings”. Economic Journal 38, 543–559. Romer, P. (1986). “Increasing returns and long-run growth”. Journal of Political Economy 94, 1002–1037. Romer, P. (1990). “Endogenous technical change”. Journal of Political Economy 98, 71–102. Segerstrom, P., Anant, T., Dinopoulos, E. (1990). “A Schumpeterian model of the life cycle”. American Economic Review 80, 1077–1092. Solow, R. (1956). “A contribution to the theory of economic growth”. Quarterly Journal of Economics 70, 65–94. Solow, R. (1957). “Technical change and the aggregate production function”. Review of Economics and Statistics 39 (3), 312–320.
CONTENTS OF VOLUME 1A
Introduction to the Series
v
Contents of the Handbook
vii
Preface to the Handbook of Economic Growth
xi
INTRODUCTION: GROWTH IN RETROSPECT AND PROSPECT Reflections on Growth Theory ROBERT M. SOLOW Abstract Keywords
3 3 3 3
PART I: THEORIES OF ECONOMIC GROWTH Chapter 1 Neoclassical Models of Endogenous Growth: The Effects of Fiscal Policy, Innovation and Fluctuations LARRY E. JONES AND RODOLFO E. MANUELLI Abstract Keywords 1. Introduction 2. Endogenous growth: Infinite lifetimes 2.1. Growth and the Solow model 2.2. A one sector model of equilibrium growth 2.3. Fiscal policy and growth 2.4. Innovation in the neoclassical model
3. Fluctuations and growth 3.1. Introduction 3.2. Empirical evidence 3.3. Theoretical models 3.4. A simple linear endogenous growth model 3.5. Physical and human capital 3.6. The opportunity cost view 3.7. More on government spending, taxation, and growth 3.8. Quantitative effects
4. Concluding comments xvii
13 14 14 15 16 17 19 21 27 32 32 32 39 39 49 54 56 59 62
xviii
Contents of Volume 1A
Acknowledgements References Chapter 2 Growth with Quality-Improving Innovations: An Integrated Framework PHILIPPE AGHION AND PETER HOWITT Abstract Keywords 1. Introduction 2. Basic framework 2.1. 2.2. 2.3. 2.4.
A toy version of the Aghion–Howitt model A generalization Alternative formulations Comparative statics on growth
3. Linking growth to development: convergence clubs 3.1. 3.2. 3.3. 3.4.
A model of technology transfer World growth and distribution The role of financial development in convergence Concluding remark
4. Linking growth to IO: innovate to escape competition 4.1. The theory 4.2. Empirical predictions 4.3. Empirical evidence and relationship to literature 4.4. A remark on inequality and growth
5. Scale effects 5.1. Theory 5.2. Evidence 5.3. Concluding remarks
6. Linking growth to institutional change 6.1. From Schumpeter to Gerschenkron 6.2. A simple model of appropriate institutions 6.3. Appropriate education systems
7. Conclusion References Chapter 3 Horizontal Innovation in the Theory of Growth and Development GINO GANCIA AND FABRIZIO ZILIBOTTI Abstract Keywords 1. Introduction 2. Growth with expanding variety 2.1. The benchmark model
63 63
67 68 68 69 69 69 71 75 75 76 78 81 81 84 84 86 89 89 92 92 92 94 97 98 98 100 101 106 107
111 112 112 113 116 116
Contents of Volume 1A 2.2. Two variations of the benchmark model: “lab-equipment” and “labor-for intermediates” 2.3. Limited patent protection
3. Trade, growth and imitation 3.1. Scale effects, economic integration and trade 3.2. Innovation, imitation and product cycles
4. Directed technical change 4.1. Factor-biased innovation and wage inequality 4.2. Appropriate technology and development 4.3. Trade, inequality and appropriate technology
5. Complementarity in innovation 6. Financial development 7. Endogenous fluctuations 7.1. Deterministic cycles 7.2. Learning and sunspots
8. Conclusions Acknowledgements References Chapter 4 From Stagnation to Growth: Unified Growth Theory ODED GALOR Abstract Keywords 1. Introduction 2. Historical evidence 2.1. The Malthusian epoch 2.2. The Post-Malthusian Regime 2.3. The Sustained Growth Regime 2.4. The great divergence
3. The fundamental challenges 3.1. Mysteries of the growth process 3.2. The incompatibility of non-unified growth theories 3.3. Theories of the demographic transition and their empirical assessment
4. Unified growth theory 4.1. From stagnation to growth 4.2. Complementary theories
5. Unified evolutionary growth theory 5.1. Human evolution and economic development 5.2. Natural selection and the origin of economic growth 5.3. Complementary mechanisms
6. Differential takeoffs and the great divergence 6.1. Non-unified theories 6.2. Unified theories
xix
120 122 124 124 127 130 131 136 140 144 150 157 158 162 166 166 166
171 172 173 174 178 179 185 195 218 219 220 221 224 235 237 256 264 264 266 273 276 277 279
xx
Contents of Volume 1A
7. Concluding remarks Acknowledgements References Chapter 5 Poverty Traps COSTAS AZARIADIS AND JOHN STACHURSKI Abstract Keywords 1. Introduction 2. Development facts 2.1. Poverty and riches 2.2. A brief history of economic development
3. Models and definitions 3.1. Neoclassical growth with diminishing returns 3.2. Convex neoclassical growth and the data 3.3. Poverty traps: historical self-reinforcement 3.4. Poverty traps: inertial self-reinforcement
4. Empirics of poverty traps 4.1. Bimodality and convergence clubs 4.2. Testing for existence 4.3. Model calibration 4.4. Microeconomic data
5. Nonconvexities, complementarities and imperfect competition 5.1. Increasing returns and imperfect competition 5.2. The financial sector and coordination 5.3. Matching 5.4. Other studies of increasing returns
6. Credit markets, insurance and risk 6.1. Credit markets and human capital 6.2. Risk 6.3. Credit constraints and endogenous inequality
7. Institutions and organizations 7.1. Corruption and rent-seeking 7.2. Kinship systems
8. Other mechanisms 9. Conclusions 9.1. Lessons for economic policy
Acknowledgements Appendix A: A.1. Markov chains and ergodicity A.2. Remaining proofs
References
283 285 285
295 296 296 297 303 303 304 307 307 312 317 326 330 330 335 337 339 340 341 343 346 349 350 351 355 358 363 364 367 373 373 374 375 375 375 378 379
Contents of Volume 1A
Chapter 6 Institutions as a Fundamental Cause of Long-Run Growth DARON ACEMOGLU, SIMON JOHNSON AND JAMES A. ROBINSON Abstract Keywords 1. Introduction 1.1. The question 1.2. The argument 1.3. Outline
2. Fundamental causes of income differences 2.1. Three fundamental causes
3. Institutions matter 3.1. The Korean experiment 3.2. The colonial experiment
4. The Reversal of Fortune 4.1. 4.2. 4.3. 4.4. 4.5. 4.6.
The reversal among the former colonies Timing of the reversal Interpreting the reversal Economic institutions and the reversal Understanding the colonial experience Settlements, mortality and development
5. Why do institutions differ? 5.1. 5.2. 5.3. 5.4.
The efficient institutions view – the Political Coase Theorem The ideology view The incidental institutions view The social conflict view
6. Sources of inefficiencies 6.1. 6.2. 6.3. 6.4. 6.5.
Hold-up Political losers Economic losers The inseparability of efficiency and distribution Comparative statics
6.6. The colonial experience in light of the comparative statics 6.7. Reassessment of the social conflict view
7. The social conflict view in action 7.1. Labor markets 7.2. Financial markets 7.3. Regulation of prices 7.4. Political power and economic institutions
8. A theory of institutions 8.1. Sources of political power 8.2. Political power and political institutions 8.3. A theory of political institutions
xxi
385 386 387 388 388 389 396 396 397 402 404 407 407 408 412 412 414 416 417 421 422 424 425 427 428 430 432 434 436 437 438 439 439 440 441 443 445 448 448 449 451
xxii
Contents of Volume 1A
9. The theory in action 9.1. Rise of constitutional monarchy and economic growth in early modern Europe 9.2. Summary 9.3. Rise of electoral democracy in Britain 9.4. Summary
10. Future avenues Acknowledgements References Chapter 7 Growth Theory through the Lens of Development Economics ABHIJIT V. BANERJEE AND ESTHER DUFLO Abstract Keywords 1. Introduction: neo-classical growth theory 1.1. The aggregate production function 1.2. The logic of convergence
2. Rates of return and investment rates in poor countries 2.1. Are returns higher in poor countries? 2.2. Investment rates in poor countries
3. Understanding rates of return and investment rates in poor countries: aggregative approaches 3.1. 3.2. 3.3. 3.4.
Access to technology and the productivity gap Human capital externalities Coordination failure Taking stock
4. Understanding rates of return and investment rates in poor countries: nonaggregative approaches 4.1. 4.2. 4.3. 4.4.
Government failure The role of credit constraints Problems in the insurance markets Local externalities
4.5. The family: incomplete contracts within and across generations 4.6. Behavioral issues
5. Calibrating the impact of the misallocation of capital 5.1. A model with diminishing returns 5.2. A model with fixed costs
6. Towards a non-aggregative growth theory 6.1. An illustration 6.2. Can we take this model to the data? 6.3. Where do we go from here?
Acknowledgements References
452 452 457 458 462 463 464 464
473 474 474 475 475 477 479 479 493 499 499 501 503 504 505 505 509 512 515 518 520 522 523 527 535 535 538 542 544 544
Contents of Volume 1A
xxiii
PART II: EMPIRICS OF ECONOMIC GROWTH Chapter 8 Growth Econometrics STEVEN N. DURLAUF, PAUL A. JOHNSON AND JONATHAN R.W. TEMPLE Abstract Keywords 1. Introduction 2. Stylized facts 2.1. 2.2. 2.3. 2.4. 2.5. 2.6. 2.7. 2.8. 2.9. 2.10.
A long-run view Data after 1960 Differences in levels of GDP per worker Growth miracles and disasters Convergence? The growth slowdown Does past growth predict future growth? Growth differences by development level and geographic region Stagnation and output volatility A summary of the stylized facts
3. Cross-country growth regressions: from theory to empirics 3.1. Growth dynamics: basic ideas 3.2. Cross-country growth regressions 3.3. Interpreting errors in growth regressions
4. The convergence hypothesis 4.1. Convergence and initial conditions 4.2. β-convergence 4.3. Distributional approaches to convergence 4.4. Time series approaches to convergence 4.5. Sources of convergence or divergence
5. Statistical models of the growth process 5.1. Specifying explanatory variables in growth regressions 5.2. Parameter heterogeneity 5.3. Nonlinearity and multiple regimes
6. Econometric issues I: Alternative data structures 6.1. 6.2. 6.3. 6.4.
Time series approaches Panel data Event study approaches Endogeneity and instrumental variables
7. Econometric issues II: Data and error properties 7.1. 7.2. 7.3. 7.4.
Outliers Measurement error Missing data Heteroskedasticity
555 556 557 558 561 562 562 563 565 567 567 568 571 573 575 576 576 578 581 582 582 585 592 599 604 607 608 616 619 624 624 627 636 637 640 641 641 642 643
xxiv
Contents of Volume 1A
7.5. Cross-section error correlation
8. Conclusions: The future of growth econometrics Acknowledgements Appendix A: Data Key to the 102 countries Extrapolation
Appendix B: Variables in cross-country growth regressions Appendix C: Instrumental variables for Solow growth determinants Appendix D: Instrumental variables for non-Solow growth determinants References Chapter 9 Accounting for Cross-Country Income Differences FRANCESCO CASELLI Abstract 1. Introduction 2. The measure of our ignorance 2.1. 2.2. 2.3. 2.4.
Basic data Basic measures of success Alternative measures used in the literature Sub-samples
3. Robustness: basic stuff 3.1. 3.2. 3.3. 3.4.
Depreciation rate Initial capital stock Education-wage profile Years of education 1
3.5. Years of education 2 3.6. Hours worked 3.7. Capital share
4. Quality of human capital 4.1. Quality of schooling: Inputs 4.2. Quality of schooling: test scores 4.3. Experience 4.4. Health 4.5. Social vs. private returns to schooling and health
5. Quality of physical capital 5.1. Composition 5.2. Vintage effects 5.3. Further problems with K
6. Sectorial differences in TFP 6.1. Industry studies 6.2. The role of agriculture 6.3. Sectorial composition and development accounting
643 645 651 651 651 652 652 660 661 663
679 680 681 683 685 686 688 689 690 690 691 693 694 694 695 696 698 698 703 706 708 710 711 711 715 716 717 718 719 724
Contents of Volume 1A
7. Non-neutral differences in technology 7.1. Basic concepts and qualitative results 7.2. Development accounting with non-neutral differences
8. Conclusions Acknowledgements References Chapter 10 Accounting for Growth in the Information Age DALE W. JORGENSON Abstract Keywords 1. The information age 1.1. Introduction 1.2. Faster, better, cheaper 1.3. Impact of information technology
2. Aggregate growth accounting 2.1. The role of information technology 2.2. The American growth resurgence 2.3. Demise of traditional growth accounting
3. International comparisons 3.1. 3.2. 3.3. 3.4. 3.5.
Introduction Investment and total factor productivity Investment in information technology Alternative approaches Conclusions
4. Economics on internet time Acknowledgements References Further reading Chapter 11 Externalities and Growth PETER J. KLENOW AND ANDRÉS RODRÍGUEZ-CLARE Abstract Keywords 1. Introduction 2. A brief guide to externalities in growth models 2.1. 2.2. 2.3. 2.4.
Models with knowledge externalities Models with knowledge externalities and new-good externalities Models with new-good externalities Models with no externalities
3. Cross-country evidence
xxv
727 727 734 737 738 738
743 744 745 746 746 747 755 756 756 765 779 784 784 787 796 803 805 806 807 808 815
817 818 818 819 819 820 821 822 823 825
xxvi 3.1. 3.2. 3.3. 3.4.
Contents of Volume 1A The world-wide growth slowdown Beta convergence in the OECD Low persistence of growth rate differences Investment rates and growth vs. levels
3.5. R&D and TFP
4. Models with common growth driven by international knowledge spillovers 4.1. R&D investment and relative productivity 4.2. Modeling growth in the world technology frontier 4.3. Determinants of R&D investment 4.4. Calibration 4.5. The benefits of engagement 4.6. Discussion of main results
5. Conclusion Acknowledgements Appendix A: Comparative statics
References
825 827 829 831 833 835 836 839 843 845 854 856 856 857 857 858 859
PART III: GROWTH POLICIES AND MECHANISMS Chapter 12 Finance and Growth: Theory and Evidence ROSS LEVINE Abstract Keywords 1. Introduction 2. Financial development and economic growth: Theory 2.1. 2.2. 2.3. 2.4. 2.5.
What is financial development? Producing information and allocating capital Monitoring firms and exerting corporate governance Risk amelioration Pooling of savings
2.6. Easing exchange 2.7. The theoretical case for a bank-based system 2.8. The theoretical case for a market-based system 2.9. Countervailing views to bank-based vs. market-based debate 2.10.Finance, income distribution, and poverty
3. Evidence on finance and growth 3.1. 3.2. 3.3. 3.4. 3.5.
Cross-country studies of finance and growth Panel, time-series, and case-studies of finance and growth Industry and firm level studies of finance and growth Are bank- or market-based systems better? Evidence Finance, income distribution, and poverty alleviation: evidence
865 866 866 867 869 869 870 872 875 879 880 881 883 886 887 888 889 899 910 918 920
Contents of Volume 1A
xxvii
4. Conclusions Acknowledgements References
921 923 923
Chapter 13 Human Capital and Technology Diffusion JESS BENHABIB AND MARK M. SPIEGEL Abstract Keywords 1. Introduction 2. Variations on the Nelson–Phelps model 3. Some microfoundations based on the diffusion model of Barro and Sala-iMartin 4. A nested specification 5. Empirical evidence 5.1. Measurement of total factor productivity 5.2. Model specification 5.3. Results
6. Model prediction 6.1. Model forecasting 6.2. Negative catch-up countries
7. Conclusion References Chapter 14 Growth Strategies DANI RODRIK Abstract Keywords 1. Introduction 2. What we know that (possibly) ain’t so 3. The indeterminate mapping from economic principles to institutional arrangements 4. Back to the real world 4.1. In practice, growth spurts are associated with a narrow range of policy reforms 4.2. The policy reforms that are associated with these growth transitions typically combine elements of orthodoxy with unorthodox institutional practices 4.3. Institutional innovations do not travel well 4.4. Sustaining growth is more difficult than igniting it, and requires more extensive institutional reform
5. A two-pronged growth strategy 5.1. An investment strategy to kick-start growth 5.2. An institution building strategy to sustain growth
935 936 936 937 940 944 946 948 948 953 954 959 959 960 964 965
967 968 968 969 973 978 989 989 993 994 996 997 998 1005
xxviii
Contents of Volume 1A
6. Concluding remarks Acknowledgements References
1009 1010 1010
Chapter 15 National Policies and Economic Growth: A Reappraisal WILLIAM EASTERLY Abstract Keywords 1. Theoretical models that predict strong policy effects 2. Models that predict small policy effects on growth 3. Empirics 4. Some empirical caveats 5. New empirical work 6. Policy episodes and transition paths 7. Institutions versus policies 8. Conclusions References
1015 1016 1016 1017 1026 1032 1033 1036 1050 1054 1056 1056
Author Index
I-1
Subject Index
I-37
REFLECTIONS ON GROWTH THEORY ROBERT M. SOLOW
Abstract This note contains some general and idiosyncratic reflections on the current state of neoclassical growth theory. It expresses some surprise at the lack of attention both to multi-sector growth models and to multi-country models with trade and capital flows. It also suggests that there might be value in further analysis of some old topics like capital–labor substitution with an expanded definition of capital, and the interaction of growth and medium-run phenomena (or, to put it differently, the interaction of demandside and supply-side variations).
Keywords economic growth, neoclassical growth model JEL classification: 041
4
R.M. Solow
I cannot remember what words Charles Dickens put in the mouth of The Ghost of Christmas Past. This is not the Cratchit family dinner anyway, and there is no one to play the role of Scrooge. But there is no doubt that I am here in roughly the capacity of the Ghost. So I will make up something for him to say. We are nearing the 50th anniversary of the neoclassical model of growth; astonishingly, it is still alive and well. There is not really any competing model. In the broad sense in which I use the term, the “endogenous growth” models of Romer and Lucas and their many successors are entirely neoclassical. So the basic model has survived for 50 years. I emphasize “basic” because progress, in theory and in practical analysis, has come mainly from extending the basic model at the edges. The territory of growth theory has expanded to include more topics in what used to be border areas. This is not necessarily exactly the same thing as “endogenizing” these borderline topics. There is a lot in the Handbook about the influence of background forces like “institutions” on the evolution of technology or total factor productivity. Some of it is in the mood of the “New Growth Theory” but not all of it. Much of it just wants to be explicit about background forces without trying to absorb them into the model. I will come back to some of the extensions of growth theory; but it is also interesting to contemplate a few of the territories into which the theory has not expanded. For example, I suspect that early on one would have expected much more work on multi-sector growth models than there has been. Not that there has been none: Leif Johansen had an early book, oriented toward planning. Luigi Pasinetti has written extensively on the sorts of structural changes to be expected along a trajectory, arising from such inevitable factors as differing income elasticities of demand for different goods. In a very different vein, there was a whole literature stemming from the von Neumann model, which now seems to have gone out of favor. Xavier Sala-i-Martin’s chapter in the Handbook reviews some worthwhile developments and promises others. In the early stages there was active exploration of two-sector models, culminating in the book by Duncan Foley and Miguel Sidrauski, but it petered out fairly soon. The reason was probably internal-intellectual rather than any feeling that the applications were unimportant. The usual, perfectly reasonable, choice was to distinguish between a consumer-good-producing sector and an investment-good sector. (Agriculture and Industry was another possible split, but mainly in the development context.) I have the feeling that too much in those models turned out to depend on differences in factorintensity between the sectors. We have very little in the way of facts or intuition about that issue, and there was no reason to expect or postulate any systematic pattern that could lead to exciting results. It is also a little odd that there was not more in the way of open-economy growth modelling. There was of course the well-known book by Gene Grossman and Elhanan Helpman; it attracted attention more for its analysis of endogenous technology and quality ladders than for trade and capital flows. Jaume Ventura’s chapter in the Handbook records the current state of play. I can only say that 40 years ago I would have been expecting to see more research in these areas than actually turned up.
Reflections on Growth Theory
5
These undercultivated subfields would have had, could still have, important practical applications. I will just mention three examples. (a) The creation and enlargement of the European Union have been modelled in several places; but I do not think that we have had the insights that could come from embedding this question in a formal multi-sector, multi-country growth model. (How much does it matter where the R&D is done? Are there potential gains from more migration within the EU? And so on.) (b) Is there anything deeper to be learned from the fact that, after a couple of decades of catch-up to the U.S. in productivity and TFP, the large European economies seem recently to have stagnated relatively or even fallen back a bit? (c) If the U.S. (and the EU) were to impose their own local environmental standards on their poor-country trading partners, what could be expected to happen to factor prices, real income levels and growth rates in the poor countries? All these things have been thought about, of course, but something might have been gained had appropriate growth models been easily available. A slightly larger and slightly different question has to do with China: How should growth theory be applied to such a large, diverse, almost dual economy, especially in a world of rapidly increasing trade and international investment? Certainly the manyindustry, many-country aspects must matter, but sheer geographical size and regional diversity may require special treatment. Chinese economists have already started applying modern growth theory to their problems; but some departures from the usual might be in order. While the Ghost is going on about might-have-beens, I will allow him a couple of paragraphs on a topic that he has grumbled about before. Neoclassical growth theory is about the evolution of potential output. In other words, the model takes it for granted that aggregate output is limited on the supply side, not by shortages (or excesses) of effective demand. Short-run macroeconomics, on the other hand, is mostly about the gap between potential and actual output. (There is an important modern school of macroeconomics that assumes this distinction away, and makes the growth model explain short-run fluctuations too. It would be a digression to discuss that issue here.) On the older view – this is after all the Ghost talking – some sort of endogenous knitting-together of the fluctuations and growth contexts is needed, and not only for the sake of neatness: the short run and its uncertainties affect the long run through the volume of investment and research expenditure, for instance, and the growth forces in the economy probably influence the frequency and amplitude of short-run fluctuations. This terrain is sometimes described as the economics of the medium run. It too has been undercultivated by growth theory. It has not been entirely ignored; but I have the impression that growth theorists simply write this off as a trivial perturbation that can not be allowed to deflect their own preoccupation with steady-state growth. For example, the work of Robert Coen and Bert Hickman, who actually do try to embed a serious demand side in a serious growth-model framework, and implement the result econometrically, is generally ignored by growth theory. There must be other scattered forays in this direction; I have taken a casual step or two myself. It should be a more
6
R.M. Solow
central part of growth theory proper. To put it differently, it would be a good thing if there were a unified macroeconomics capable of dealing with trend and fluctuations, rather than a short-run branch and a long-run branch operating under quite different rules. My quarrel with the real business cycle and related schools is not about that; it is about whether they have chosen an utterly implausible unifying device. I can now turn from the things that growth theory has not accomplished to the things that it has done, in particular the way it has expanded outside the confines of a narrow model. The main effort has quite properly gone into the endogenization of changes in technology (or more broadly TFP, though usually with technology in mind) and changes in the stock of human capital. In both cases the popular early theoretical models had features that I personally found unappealing and, in policy terms, misleading for reasons that I have pursued elsewhere, and do not intend to repeat now. On the whole, better ideas have driven out worse ones as they are supposed to do. Both lines of research – technology and human capital – have led to a welcome emphasis on social norms and institutions as enabling or limiting factors or even as actual sources of growth. The extent of interest in such ideas is represented explicitly in the Handbook by the chapters by Acemoglu–Johnson–Robinson, Greif, Alesina, Parente–Prescott, and implicitly by others. This emphasis on the role of institutions at least opens up the possibility – about which I am now more optimistic than I once was – of connecting up growth theory with the problem of economic development, in which issues of institutional change are clearly central. My own prejudice – Ghosts are allowed, even encouraged, to have prejudices – is that there may have been a premature tendency to assimilate growth and development, abetted by the vogue for cross-country regressions. A country is a country, one might say, just another point in (n + 1)-dimensional space, although loud squeals from the data have sometimes forced a restriction of the sample to OECD countries. This is something that needs to be straightened out; and detailed analysis of institutions is probably a better method than cross-country regressions. The breathtakingly broad sweep of the story-line proposed by Daron Acemoglu and colleagues is irresistibly fascinating. Much of it has the ring of truth. I must confess nevertheless to a certain skepticism about firm conclusions at this level of generality, especially when they bear on “ultimate” causality. “Good” political institutions can certainly make the path to growth-friendly economic institutions shorter and smoother. But there are cases of “bad” – autocratic – governments opting for enforceable property rights and other “good” economic institutions, possibly in the belief that economic success will ultimately strengthen the hand of the autocrats themselves. Singapore and early post-war South Korea are examples; one of them evolved toward political democracy and the other did not. The interaction of political institutions and the available stock of human capital can be very complex. Very poor countries are usually characterized by very bad governments and very deficient human capital, and these probably reinforce one another. I would not find it hard to accept the notion that there is no universally reliable way to escape this trap. Some countries succeed and others fail, for reasons that may be totally obscure ex ante and only partially and tentatively explicable ex post.
Reflections on Growth Theory
7
Whatever generalizations we are prepared to accept, however, there has to be a next stage: after history has made it plain that secure property rights and markets are better for growth than mere hierarchical rent-extraction, what do you do for an encore? The devil is in the details. What sort of patent protection provides the best mixture of incentives for innovation and diffusion? How should the free-rider problem intrinsic to non-firm-specific training best be handled? Do alternative feasible norms for corporate governance have any significant implications for growth? These and many other institutional choices are practically invisible on the Acemoglu scale, but they bulk pretty large if you are considering alternative policies for a growing capitalist economy or for a transition economy. A very similar point is made by Rodrik in his chapter of the Handbook, but in a slightly different context. It is reassuring that many of the same considerations that preoccupy Acemoglu et al. also figure at the nittygritty level: those institutional choices have real distributional consequences that can in turn help or hurt the vested interests that in turn may be able to block those very choices. Can the protagonists be bought off or overcome politically or satisfied by compromise? The familiar elasticities and marginal whatnots come back into play as soon as one tries to face up to those questions in systematic but practical ways. The translation of any “institutional” question into the language of an aggregative model is always tricky. The concepts and quantities that appear in an economic model need not be capable of expressing what a knowledgeable observer would like to say about institutional differences. An example of this occurs in the pleasing and informative chapter by Philippe Aghion and Peter Howitt. It is an incidental matter that caught my eye because it relates to some independent work of my own. The standard Schumpeterian wisdom is that active competition is bad for innovation because it erodes entrepreneurial rents too soon. But the standard empirical–historical finding is that active competition is associated with productive innovation. Aghion and Howitt find a way to enlarge their model so that the competition-innovation nexus can in principle go either way. (It is a clever and useful device that the interested reader should study.) The trouble is that the competition that Aghion and Howitt can conveniently model is between an innovating monopolist and the competitive fringe still stuck with the older technology. But the typical way in which the absence of competition deadens economic performance is that regional or national monopolies are protected by legal or other barriers against competition from best-practice firms, or just better-practice firms, domestic or foreign. They can survive without innovating or adopting best practice and they do so. To incorporate that chain of causation in a standard growth model might be difficult. Is it worth doing? We will not know until somebody tries. No one would claim that we now have a really good causal account of either technological and organizational progress or of the accumulation of human capital in all of its various forms. (Schooling is not the same thing as human capital.) As the Handbook shows, a lot of ingenuity has gone into that research and lot of ground has been covered; if there is a lot more ground still to be explored, that is hardly surprising. Nor is there any guarantee that everything one would want to know about these processes
8
R.M. Solow
is knowable. Some regression residuals represent not omitted variables but mere sound and fury, signifying nothing. It is possible that the obvious importance and interest of these matters, combined with a less worthy temptation, the sheer convenience of the Cobb–Douglas production function, may have diverted attention from an older-fashioned topic, namely the substitution of (physical and human) capital for raw labor. In the beginning, one of the surprising implications of the neoclassical growth model was not merely that the steady-state growth rate was independent of the saving-investment rate, but perhaps even more that the (moving) equilibrium level of output per person apparently responds very weakly to changes in the saving-investment quota. To be more precise, the elasticity of output per worker with respect to the saving-investment rate is the ratio of the capital share to the labor share (in steady-state equilibrium). Back then the conventional numbers were 1/4 and 3/4, giving an elasticity of 1/3. The message appeared to be that as big an increase in the rate of investment as policy could manage would yield only a disappointing increase in productivity. The conventional numbers have changed, partly as a matter of fact and partly because the capital concept has been enlarged. Apart from that, there may be some interesting points of principle to be investigated. (I have to admit that I dwell on this because Olivier de La Grandville and I have written a substantial paper on the subject.) The point I want to make here starts from the fact that the relative shares of capital and labor once seemed to be trendless in modern industrial economies; and that may still be a fair description. (This generally accepted fact provided a respectable justification for the addiction of theorists to Cobb–Douglas.) Nowadays, however, there is some basis for thinking that the capital share may have risen. It is useful in this connection to keep in mind that the aggregative elasticity of substitution, taken as measuring the responsiveness of relative factor prices to relative factor intensity, is not a purely technological concept (as John Hicks realized when he invented it in 1932). One quantitatively important way in which the aggregate economy can substitute capital for labor is through a factor-price-and-commodity-price-induced shift in demand from labor-intensive to capital-intensive goods and services. That sort of easy substitution, just like easy technical substitution, is also a way of fending off diminishing returns. So it is at least thinkable that the aggregative elasticity of substitution might be or might become fairly large, especially at high capital–labor ratios. It was part of the original neoclassical growth model that a large enough elasticity of substitution allows sustained growth in output per worker and capital per worker even without technological progress (and faster still with it). Now, to get back to where I started, since the elasticity of substitution is larger than one in such a process, the (competitive) share of capital would be rising along with the capital–labor ratio, indeed rising toward one. But then the elasticity of productivity with respect to the saving-investment rate would be getting very large. Remember: I am not prepared to tell this story, only to suggest that the mechanics of the aggregative elasticity of substitution might be as interesting an object of study for growth theorists as the last
Reflections on Growth Theory
9
little twist on endogenous technical progress. There may be several ways to postpone or hold off the influence of diminishing returns to broadly-defined capital. Discussion of these various facets of growth theory calls to mind a background decision that is rarely explicit: What sort of characteristic time interval are we talking about? Is growth theory 5-year, 20-year or 100-year economics? I know what I had in mind at the beginning: growth models were about intervals of time just long enough so that deviations above or below potential output would be small relative to the increment to output from beginning to end. An older terminology would have said that the model is about trend with business cycle removed; but that is awkward because the model itself is supposed to determine the trend, and I have doubts about the utility of the “cycle” idea. Anyhow that was then. The range of time perspectives implicit in current research, as represented in the Handbook, is very broad. At one extreme, many macroeconomists propose to use a growth model to describe events quarter by quarter, presumably week by week if the data were available. At the other extreme, the focus on political economy and institutional change seems to call for marking the clock in intervals of a century more or less. I have already suggested that I think the first choice seems willfully to ignore or deny economic events that ought to be important in macroeconomics. The very long perspective tends to treat as mere perturbations or disturbances changes that ought to be at the center of growth theory itself. If I were using a growth model to interpret century-long time series, I would want to re-calibrate the parameters a couple of times in between if the data seemed to call for it. I think I am driven back to the earlier convention. To be excessively concrete: an economy growing at an average rate of three percent a year will roughly double in 25 years. Episodes of recession or overheating amounting to five percent of total output would cover most short-run fluctuations, and would probably count as relatively small compared with overall growth. This suggests to me that the natural habitat of growththeoretic explanations is time-spans of 25 to perhaps 40 or 50 years. Anything much shorter is likely to mix up supply-side and demand-side effects, and anything much longer runs the risk of overlooking some events that ought to be accounted for explicitly. That cannot be called an error; the very-long-run perspective has its own rewards. But then we needs some nomenclature to distinguish it from the 30-year perspective. These scattered observations and remarks seem to coalesce – at least they do for me – in a general reflection. I have set it down before, obscurely, and it may bear repeating. Growth theory has focused mainly on the steady state. This is partly because accounting for Nicholas Kaldor’s “stylized facts” was one of the early goals of growth modelling; and they were essentially a description of a steady state. In addition, it is much easier to work out general and robust properties of steady states than to do the same for transition paths. So the emphasis on steady states is both explicable and reasonable. There is a side-effect, however, that strikes me as not so good. Somehow the convention has become established that a policy aimed at “growth” is by definition a policy that will increase the steady-state growth rate. A policy package that merely increases output by 10 percent at every point along an already established trajectory is somehow disap-
10
R.M. Solow
pointing. You might take this as just another example of the principle that where there is life there is hype. But for growth theorists it has led to a premium on models that do offer a direct connection between easily manageable policies and the steady-state growth rate. The early crop of so-called “AK” models was the predictable response. I thought they were uninteresting theory, in the sense that they more or less assumed what they purported to prove, and also misleading guides to policy, in that they made something look easy that is in fact very difficult. The fashion for such models seems to have waned. William Easterly’s chapter in the Handbook demonstrates rather convincingly that these apparently easy levers on the growth rate are indeed illusory, as we should have known. If you look closely, however, even more serious models of endogenous growth seem to depend, at a key point, on the blunt assumption that, for some important X, dX/dt = A(. . .)X, where A(. . .) is a function of one or more easily manipulable level variables. Maybe that is a logical necessity: if you want an exponential solution, you better have a linear differential equation. One has to accept that piece of wisdom. The real point, however, is that any such linearity assumption, because it is so powerful, ought to require much more convincing justification than it gets in the standard models of endogenous technological change or accumulation of human capital. I wonder if it would be a fair summary of the status quo to say that the broad neoclassical model of growth is widely accepted as a valid description of the mechanism of economic growth in advanced economies; most current research is aimed at unpacking, understanding and testing those aspects that the basic model takes as given. The Ghost could go along with that, subject to a couple of minor amendments already stated. The mechanism itself needs extension so that it can cover international flows of goods, capital and technology (and perhaps labor), so that it can better describe interactions between demand-driven fluctuations and the path of potential output, and so that it can allow explicitly for the existence of many goods and sectors with different technologies and different demand conditions. These needs were foreseeable and foreseen early in the story; they are merely unfinished old business.
Chapter 1
NEOCLASSICAL MODELS OF ENDOGENOUS GROWTH: THE EFFECTS OF FISCAL POLICY, INNOVATION AND FLUCTUATIONS LARRY E. JONES University of Minnesota and Federal Reserve Bank of Minneapolis RODOLFO E. MANUELLI University of Wisconsin
Contents Abstract Keywords 1. Introduction 2. Endogenous growth: Infinite lifetimes 2.1. Growth and the Solow model 2.2. A one sector model of equilibrium growth 2.3. Fiscal policy and growth 2.3.1. Quantitative analysis of the effects of taxes 2.3.2. Productive government spending 2.4. Innovation in the neoclassical model 2.4.1. Notation 2.4.2. Balanced growth properties of the model 2.4.3. Adding a non-convexity
3. Fluctuations and growth 3.1. Introduction 3.2. Empirical evidence 3.3. Theoretical models 3.4. A simple linear endogenous growth model 3.4.1. Case 1: An Ak model 3.4.2. Case 2: A two sector (technology) model 3.4.3. Case 3: Aggregate vs. sectoral shocks 3.5. Physical and human capital 3.6. The opportunity cost view 3.7. More on government spending, taxation, and growth
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01001-4
14 14 15 16 17 19 21 23 25 27 28 30 31 32 32 32 39 39 43 46 47 49 54 56
14
L.E. Jones and R.E. Manuelli 3.8. Quantitative effects
4. Concluding comments Acknowledgements References
59 62 63 63
Abstract Despite its role as the centerpiece of modern growth theory, the Solow model is decidedly silent on some of its basic questions: Why is average growth in per capita income so much higher now than it was 200 years ago? Why is per capita income so much higher in the member countries of the OECD than in the less developed countries (LDC) of the world? In this chapter we review the recent literature on endogenous growth. We concentrate on convex models and we restrict attention to the case in which markets are competitive. After a brief review of the basic mechanisms that produces growth, we concentrate on three topics: the impact of fiscal policies on growth, the role of innovation and the relationship between uncertainty and growth.
Keywords endogenous growth, convex models, competitive markets, taxation, innovation, uncertainty JEL classification: H2, H3, O4
Ch. 1: Neoclassical Models of Endogenous Growth
15
1. Introduction Despite its role as the centerpiece of modern growth theory, the Solow model is decidedly silent on some of its basic questions: Why is average growth in per capita income so much higher now than it was 200 years ago? Why is per capita income so much higher in the member countries of the OECD than in the less developed countries (LDC) of the world? The standard implementation of the Solow model provides no answers for these questions except, perhaps, for differences across time and across countries in the production possibility set. This is typically summarized by differences in Total Factor Productivity (TFP). The fundamental reasons for why TFP might be different in different countries, or in different time periods is left open for speculation. If these differences are supposed to be due to differences in innovations, it is not made clear why access to these innovations should be different, nor is it noted that these innovations themselves are economic decisions – they have costs and benefits, and are made by optimizing, private agents. This basic weakness in the Solow model (and its followers) was the driving force behind the development of the class of endogenous growth models. This literature has been wide and varied, with the models developed ranging from perfectly competitive, convex models to ones featuring a range of types of market failures (e.g., increasing returns, external effects, imperfectly competitive behavior by firms, etc.). A common feature that has been emphasized throughout is knowledge, or human capital, and its production and dissemination. In some cases, this has been directly treated in the modeling, in others it has been more tangential, an important consideration for quantitative development, but less so for qualitative work. That this focus is essential follows from the fact that the Solow model already accurately reflects the quantitative limits of using models with only physical capital. (That is, capital’s share is determined by the data to put us in the Solow range, technologically.) Although they differ in their details, in the end, what this class of models points to as differences in development are differences in social institutions across time and countries. Thus, countries that have weaker systems of property rights, or higher wasteful taxation and spending policies, will tend to grow more slowly. Moreover, these differences in performance can be permanent if these institutions are unchanging. As a corollary, those countries who developed these growth enhancing institutions more recently (and some still have not), have levels of income that are lower than those in which they were adopted earlier, even if current growth rates show only small differences. In this paper we limit ourselves to studying neoclassical models. By this we mean models with convex production sets, well behaved preferences and a market structure that is consistent with competitive behavior. Therefore, we do not review the large literature that addresses the role of externalities and non-competitive markets. As it turns out, most of the basic ideas behind this literature can be expressed in simple, convex models of aggregate variables without uncertainty. These are the models that are the first focus of this chapter. They have proved both highly flexible and easy to use. With them, we can give substance to statements like those above that property rights and other govern-
16
L.E. Jones and R.E. Manuelli
mental institutions are key to long run growth rates in a society. Most of this branch of the literature is well known by now, and much of it appears on standard graduate macro reading lists. Accordingly, our discussion will be fairly brief.1 One important, and as of yet unresolved issue, is the size of the growth effects of cross-country differences in fiscal policy. Thus, our review of the standard convex model is complemented with a discussion of the more recent findings about the quantitative effects of taxes (and government spending) on growth. Even though the theoretical effects of social institutions are well understood, this is less true of the recent work on perfectly competitive models of innovation, and so, comparatively more space is used to discuss that ongoing development. As a second focus, one issue that comes immediately to light in studying this class of models is the possibility that uncertainty per se might have an impact on long run performance. This points to the possibility that instability in property rights and institutions might change the incentives for investment. That is, how are the time paths of savings, consumption and investment affected by uncertainty in this class of models? How does this compare with how uncertainty affects decisions in the Solow model [i.e., Brock and Mirman (1972) vs. Cass (1965) and Koopmans (1965)]? Much less is known about the answers to these questions at the present time and that knowledge that does exist is much less widespread. For this reason, we present a fairly detailed discussion of the properties of stochastic, convex models of endogenous growth. To this end, we study models in which technologies and policies are subject to random shocks. We characterize the effects of differential amounts of uncertainty on average growth. We show that increased uncertainty can increase or decrease average growth depending on both the parameters of the model and the source of the uncertainty. A separate, but related topic, is the business cycle frequency properties of these models. This is left to future work. In Section 2, we lay out the basics of the class of neoclassical (i.e., convex) models of endogenous growth. We show how differences in social institutions across time and across countries can give rise to different performance, even over the very long run. We also lay out some of the interpretations of the model, including human capital investment and innovation and knowledge diffusion sectors, that lend richness to its interpretation. Section 3 discusses properties of the models when uncertainty is added, and shows how this can affect the long run growth rate of an economy.
2. Endogenous growth: Infinite lifetimes Historically, the engine of growth as depicted in Solow’s seminal work on the topic (1956) was the assumption of exogenous technical change. Thus, initially, growth models aimed at being consistent with growth facts, but gave up on the possibility of 1 Other authors have also presented comprehensive surveys of this literature [see Barro and Sala-i-Martin (1995), Jones and Manuelli (1997), and Aghion and Howitt (1998) for examples]. Our aim is to complement those presentations, rather than repeat them, and hence, our focus is somewhat distinct.
Ch. 1: Neoclassical Models of Endogenous Growth
17
explaining them. Moreover, this approach has weaknesses in two distinct areas. First, it is difficult using the exogenous growth model to explain the observed long run differences in performance exhibited by different countries. Second, the productivity changes that are assumed exogenous in the Solow model are, in fact, the result of conscious decisions on the part of economic agents. If this is the case, it is then important to explore both the mechanism through which productivity changes as well as the factors that can give rise to the observed long run differences if we are to understand these phenomena. In this section we briefly review the basic optimal growth model as initially analyzed by Cass (1965) and Koopmans (1965). We then discuss the nature of the technologies consistent with endogenous growth and the role of fiscal policy in influencing the growth rate. We conclude with an analysis of the role of innovation in the context of convex models of equilibrium growth. 2.1. Growth and the Solow model In the simplest time invariant version of the Solow model, it can be shown that the per capita stock of capital converges to a unique value independent of initial conditions. It is then necessary to assume some exogenous source of productivity growth in order to account for long run growth. In Solow (1956), it is assumed that labor productivity grows continually and exogenously. In response, the capital stock (assumed homogeneous over time) is continually increased allowing for a continual expansion in the level of output and consumption. The literature on endogenous growth has concentrated on replacing this assumed exogenous productivity growth by an endogenous process. If this change in productivity of labor is thought to arise from the invention of techniques consciously developed, the literature on endogenous growth can then be thought of as explicitly modeling the decisions to create this technological improvement [see Shell (1967) and (1973)]. For this to go beyond a reinterpretation of the Solow treatment, it must be that the technology for discovering and developing these new technologies does not have itself a source of exogenous technological change. Because of this, these models all feature technologies that are time stationary. The consumer problem in the simple growth model is given by max {ct }
∞
β t u(ct )
t=0
subject to ∞
pt (ct + xt ) W0 +
t=0
kt+1 = (1 − δk )kt + xt ,
∞
pt rt kt ,
(1a)
t=0
(1b)
where ct is the level of consumption, xt is investment, kt is the capital stock, pt is the price of consumption (relative to time 0), and rt is the rental price of capital, all in
18
L.E. Jones and R.E. Manuelli
period t, and W0 is the present value of wealth net of capital income. The first order condition for (an interior) solution to this problem is just u (ct ) = βu (ct+1 )[1 − δk + rt+1 ].
(2)
If, as is standard in the literature, the instantaneous utility function, u(ct ), is assumed strictly concave, growth – defined as a situation in which ct+1 > ct – requires β[1 − δk + rt+1 ] > 1.
(3)
Condition (3) is fairly general, and must hold independently of the details of the production side of the economy. Thus, if the economy is going to display long run growth, the rate of return on savings must be sufficiently high. What determines the economy’s rate of return? In the standard Solow growth model – and in many convex models – firms can be viewed as solving a static problem. More precisely, each firm maximizes profits given by Πt = max c + x − rt k − wt n, k,n
subject to c + x F (k, n), where F is a concave production function that displays constant returns to scale. Since in equilibrium the household offers inelastically one unit of labor, the rental rate of capital must satisfy rt = f (kt ),
(4)
where f (k) = F (k, 1), and k is capital per worker. It is now straightforward to analyze growth in the Solow model. The equilibrium version of (2) is just u (ct ) = βu (ct+1 ) 1 − δk + f (kt+1 ) . (5) If the productivity of capital is sufficiently low as the stock of capital per worker increases, then there is no long run growth. To see this, note that if limk→∞ f (k) = r, with 1 − δk + r < 1, there exists a finite k ∗ such that 1 − δk + f (k ∗ ) = 1. It is standard to show that the unique competitive equilibrium for this economy (as well as the symmetric optimal allocation) is such that the sequence of capital stocks {kt } converges to k ∗ . Given this, consumption is also bounded. (Actually, it converges to f (k ∗ ) − δk k ∗ .) Can exogenous technological change ‘solve’ the problem. The answer depends on the nature of the questions that the model is designed to answer. If one is content to generate equilibrium growth, then the answer is a clear yes. If, on the other hand, the objective is to understand how policies and institutions affect growth, then the answer is negative.
Ch. 1: Neoclassical Models of Endogenous Growth
19
To see this assume that technological progress is labor augmenting. Specifically, assume that, at time t, the amount of effective labor is zt = z(1+γ )t . In order to guarantee existence of a balanced growth path we assume that the utility function is isoelastic [see Jones and Manuelli (1990) for details], and given by u(c) = c1−θ /(1 − θ ). Let a ˆ over a variable denote its value relative to effective labor. Thus, cˆt ≡ ct /(z(1 + γ )t ). In this case, the balanced growth version of (2) is ˆ (cˆt+1 ) 1 − δk + f kˆt+1 , (1 + γ )u (cˆt ) = βu where βˆ = β(1 + γ )1−θ .2 Standard arguments show that the equilibrium of this economy converges to a steady ˆ Thus, this implies that, asymptotically, consumption is given by ct = (1 + state (c, ˆ k). t γ ) zc. ˆ Thus, even though there is equilibrium growth, the growth rate is completely determined by the exogenous increase in labor augmenting productivity. 2.2. A one sector model of equilibrium growth As we argued before, the critical assumption that results in the economy not growing is that the marginal product of capital is low. The modern growth literature has emphasized the analysis of economies in which the marginal product of capital remains (sufficiently) bounded away from zero. This induces positive long-run growth in equilibrium. As we will show, how fast output grows in these models depends on a variety of factors (e.g., parameters of preferences). Because of this, these models have the property that the rate of growth is determined by the agents in the model. Throughout, there will be one common theme. This mirrors the point emphasized above, that for growth to occur, the interest rate (either implicit in a planning problem or explicit in an equilibrium condition) must be kept from being driven too low. This follows immediately from the discussion above. In terms of key features of the environment that are necessary to obtain endogenous growth there is one that stands out: it is necessary that the marginal product of some augmentable input be bounded strictly away from zero in the production of some augmentable input which can be used to produce consumption. Since we are dealing with convex economies, the arguments in Debreu (1954) apply to the environments that we study. Thus, in the absence of distortionary government policies, equilibrium and optimal allocations coincide. Thus, for ease of exposition, we will limit ourselves to analyzing planner’s problems. The planner’s problem in the basic one sector growth model is given by max {ct }
∞
β t u(ct ),
t=0
2 Existence of a solution requires that β(1 + γ )1−θ < 1, which we assume.
20
L.E. Jones and R.E. Manuelli
subject to ct + xt F (kt , nt ), kt+1 (1 − δk )kt + xt , where ct is per capita consumption, kt is the per capita stock of capital, xt is the (nonnegative) flow of investment, and nt is employment at time t. Since we assume that leisure does not yields utility, the optimal (and equilibrium) level of nt equals the endowment, which we normalize to 1. The Euler equation for this problem is just (5) given that, as before, we set f (k) = F (k, 1). It follows that if limk→∞ β[1 − δk + f (k)] > 1, then lim supt ct = ∞. Thus, there is equilibrium growth. This result does not depend on the assumption of just one capital stock. More precisely, in the case of multiple capital stocks, the feasibility constraint is just ct +
I
xit f (k1t , . . . , kI t ),
i=1
kit+1 (1 − δik )kit + xit . In this case, the natural analogue of the assumption that the marginal product of capital is bounded is just that there is a homogeneous of degree one function – a linear function – that is a lower bound for the actual production function. However, it turns out that all that is required is that there exist a ray that has bounded marginal products. Formally, this corresponds to C ONDITION 1. Assume that f (k1 , . . . , kI ) h(k1 , . . . , kI ), where h is concave, homogeneous of degree one and C 1 for all (k1 , . . . , kI ) ∈ RI+ . Moreover, assume that there exists a vector kˆ = (kˆ1 , . . . , kˆI ), kˆ = 0, such that if kˆi > 0, β 1 − δk + hi kˆ > 1, i = 1, . . . , I. The basic result is the following [see Jones and Manuelli (1990)]. P ROPOSITION 2. Assume that Condition 1 is satisfied. Then, any optimal solution {ct∗ } is such that lim supt ct∗ = ∞. As Jones and Manuelli (1990) show, the planner’s solution can be supported as a competitive equilibrium. An extension to multiple goods is presented by Kaganovich (1998) and it is based on similar insights. It is clear that Condition 1 does not rule out decreasing returns to scale. This, in turn implies that this class of models is consistent with a version of the notion of conditional convergence: relatively poor countries are predicted to grow faster than richer countries, with the consequent closing of the income gap. Put it differently, theory suggests that, with a finite amount of data, it is difficult to distinguish an endogenous growth model from a Cass–Koopmans exogenous growth model. The main difference lies in the tail behavior of the relevant variables (output or consumption), and not in the balanced (or unbalanced) nature of the equilibrium path.
Ch. 1: Neoclassical Models of Endogenous Growth
21
2.3. Fiscal policy and growth In this section we describe the effects of taxes and government spending on the long run growth rate. Consider the problem faced by a representative agent max
∞
β t u(ct , 1 − nt )
t=0
subject to 1 + τ c ct + 1 + τ x pt xkt + 1 + τ h qt wt 1 − τ n (nct ht + nkt ht ) + 1 − τ k rt kt + Tt + Πt , where τ j represent tax rates, ct is consumption, xkt is investment in physical capital, qt are market goods used in the production of human capital, nit ht is effective labor – the product of human capital and hours – allocated to sector i, kt is the stock of capital, Tt is a government transfer, and Πt are net profits. Accumulation of human capital at the household level satisfies ht+1 (1 − δh )ht + F h (qt , nht ht ), where F h is homogeneous of degree one, concave and increasing in each argument. The economy has two sectors: producers of capital and consumption goods. Output of the capital goods industries satisfies xt F k (kkt , nkt ht ), where F k is homogeneous of degree one and concave. Feasibility in the consumption goods industry is given by ct F c (kct , nct ht ), where F c is increasing and concave. It is not necessary to assume that this production function displays constant returns to scale. It is illustrative to consider several special cases. Throughout, we assume that the utility function is of the form that is consistent with the existence of a balanced growth path. Specifically, we assume that u(c, ) = (cv())1−θ /(1 − θ ). Moreover, since our emphasis is on the role of taxes and tax-like wedges between marginal rates of substitution and transformation, we assume that lump sum transfers, Tt , are adjusted to satisfy the government budget constraint. Case I: One sector model with capital taxation. We assume that the consumer supplies one unit of labor inelastically. In this case F c = F k = Ak + Fˆ (k), where Fˆ (k) is strictly concave and limk→∞ Fˆ (k) = 0. For now we ignore human capital and set F h ≡ 0. It follows that the balanced growth rate satisfies 1 − τk θ γ = β 1 − δk + A . 1 + τx
22
L.E. Jones and R.E. Manuelli
Thus, in this setting, increases in the effective tax on capital, (1 − τ k )/(1 + τ x ) unambiguously decrease the equilibrium tax rate. Thus, unlike exogenous growth models, government policies affect the growth rate. Moreover, this simple example illustrates that the size of the impact of changes in tax rates on the long run growth rate depends on the intertemporal elasticity of substitution 1/θ . More precisely the elasticity of the growth rate with respect to τ k is given by τk
1 ∂γ τ k 1+τ x A =− . k ∂τ γ θ 1 − δk + 1−τ xk A 1+τ
It follows that, other things constant, high values of the intertemporal elasticity of substitution result in large changes in predicted growth rates in response to changes in tax rates. Thus, even an example as simple as this one illustrates that the quantitative predictions of this class of models will heavily depend on the values of the relevant preference (and technology) parameters. Case II: Physical and human capital: Identical technologies. In this section we assume that F c = F k , and F h = q. This implies that all three goods – investment, consumption and human capital – are produced using the same technology and, in particular, the same physical to human capital ratio. As in the previous section, τ k and τ x do not play independent roles. Thus, to simplify notation, we will set τ x = 0. However, the reader should keep in mind that increases in the tax rate on capital income are equivalent to increases in the tax rate on purchases of capital goods. In this case, the balanced growth conditions are γ θ = β 1 − δk + 1 − τ k Fk (κ, n) , (6a) 1 − τn c v (1 − n) = (6b) Fn (κ, n), h v(1 − n) 1 + τc 1 − τn 1 − τ k Fk (κ, n) − δk = (6c) Fn (κ, n)n − δh , 1 + τh c + (γ + δk − 1) = F (κ, n). (6d) h There are several interesting points. First, increases in the tax rate on consumption goods (i.e. sales or value added taxes) are equivalent to increases in the tax rate on labor income. Second, the relevant tax rate to evaluate the return on human capital is (1 − τ n )/(1 + τ h ). Thus, it is possible that increases in τ n – as observed in the U.S. between the pre World War II and the post WWII periods – if matched by decreases in τ h (corresponding, for example, to expansion in the quantity and quality of free public education) have no effect on the physical capital–human capital ratio, κ. Third, it is possible to show that increases in τ k , τ n , τ h or τ c result in lower growth rates. Last, without making additional assumptions about preferences and technology, it is not possible to sign the impact of changes in tax rates on other endogenous variables.
Ch. 1: Neoclassical Models of Endogenous Growth
23
Case III: Physical and human capital: Different factor intensities. In this case, we assume that only human capital is used in the production of human capital. Thus, F h = Ah nh h. This is the technology proposed by Uzawa (1964) and popularized in this class of models by Lucas (1988). For simplicity, we only consider capital and labor taxes. The relevant steady state conditions are (6a), (6b), and (6d). However, (6c) becomes γ θ = β[1 − δh + Ah nh ].
(7)
In this version of the model, changes in labor income taxes, reduce growth through their impact on hours worked (relative to leisure). However, if total work time is inelastically supplied, i.e. v() ≡ 1, the growth rate is pinned down by γ θ = β[1 − δh + Ah ]. Thus, in this setting [which corresponds to Lucas (1988) model without the externality, and to Lucas (1990)], taxes have no effect on growth. Increases in the tax rate on capital income simply change physical capital–human capital ratio and they leave the after tax rate of return unchanged. The reason for this extreme form of neutrality is that even though taxes on labor income reduce the returns from education, they also reduce the cost of using time to accumulate human capital (the value of time decreases with increases in taxes), and the two changes are identical. Thus, the cost-benefit ratio of investing in education is independent of the tax code. 2.3.1. Quantitative analysis of the effects of taxes Since the development of endogenous growth theory there have been several studies of the implications of substituting lump-sum taxes for a variety of distortionary taxes. Jones, Manuelli and Rossi (1993) analyze the optimal choice of distortionary taxes in several models of endogenous growth. In the case that physical and human capital are produced using the same technology and labor supply is inelastic, they find that for parameterizations that make the predictions of the model consistent with observations from the U.S., the potential growth effects of drastically reducing (eliminating in most cases) all forms of distortionary taxation is quite high. For their preferred parameterization the increase in growth rates is about 3%. They study a version of the model in which F c = F k = F h , and the functions F k and F h are both of the Cobb–Douglas variety, but differ in the average productivity of capital. Jones, Manuelli and Rossi estimate the capital share parameter to be equal 0.36 in the consumption sector, and to be somewhere in the 0.40–0.50 range in the human capital production sector.3 They also allow labor supply to be elastic. Their findings suggest that switching to an optimal tax code result in increases in yearly growth rates of somewhere between 1.5% and 2.0% per year. These are substantial effects. 3 Jones, Manuelli and Rossi (1993) calibrate this share. Since they study the sensitivity of their results to changes in other parameters (e.g., the intertemporal elasticity of substitution), the market goods share is not constant across experiments.
24
L.E. Jones and R.E. Manuelli
The third experiment that they consider involves the endogenous determination (by the planner) of the level of government consumption. In this case, they revert back to the one sector version of the model, and they explore not only the consequences of changing the intertemporal elasticity of substitution, but they allow for varying elasticity of substitution between capital and human capital. For their preferred characterization, they also find growth effects of about 2% per year. Moreover, as in the other experiments, the predictions are quite sensitive to the details of the model – in particular, to the choice of the intertemporal elasticity of substitution, and the degree of substitutability between capital and human capital. Stokey and Rebelo (1995) undertake a thorough review of several models that estimate the growth impact of tax reform. They argue that in the U.S. tax rates in the post WWII period are significantly higher than in the pre WWII era. This conclusion is based on the increase in the revenue from income taxes as a fraction of GDP in the early 1940s. To reconcile the models with this evidence, they conclude that the human capital share in the production of human capital must be large, and that this sector must be lightly taxed. This description is close to the Case III above and, as argued before, it results in no growth effects.4 Thus, in agreement with Lucas (1990) – and using a very similar specification of the human capital production technology – they conclude that changes in tax rates cannot have large growth effects. This conclusion depends on several assumptions. First, that the U.S. evidence shows an increase in the general level of taxes after WWII. Second, that even if there is a tax increase, the additional revenue is used to finance lump-sum transfers. Third, that the balanced growth path is a good description of the pre and post WWII economy. Measuring changes in the relevant marginal tax rates is a difficult task. Barro and Sahasakul (1986) using tax records compute average marginal tax rates for the U.S. economy. Their estimates, consistent with the Stokey and Rebelo assumption, show an increase in the 1940s. Even though the evidence about changes in the tax rate consistently points to an increase, the implications of this result for the model are not obvious. Consider, first, the uses of tax revenue. If, for example, additional income tax revenues (at the local level) are used to finance local publicly provided goods (e.g., education), then Tiebout-like arguments suggest that the ‘tax effect’ of a tax increase is zero. In the U.S. a substantial increase in government spending corresponds to increases in expenditures on education and, hence, the possibility of individuals sorting themselves to buy the ‘right’ bundle of publicly provided private goods cannot be ignored. A second quantitatively important public spending program in the post WWII era is Social Security. To the extent that benefits are dependent on contributions, the statutory tax rate on labor income used to finance social security overstates the true tax rate.5 In this case,
4 The results are continuous in the parameters. Thus, for market goods share close to zero, as Stokey and
Rebelo prefer, the growth effects are small. 5 In a pay-as-you-go system, even if the share of total payments that an individual receives is sensitive to his contributions, the same effect obtains.
Ch. 1: Neoclassical Models of Endogenous Growth
25
tax payments purchase the right to an annuity whose value is dependent on the payment. Finally, in a model with multiple tax rates an increase in a single tax does not imply, necessarily, a decrease in the growth rate. For the U.S. the evidence on the time path of capital income taxes is mixed. In a recent study, Mulligan (2003) argues that the tax rate on capital income has steadily fallen in the last 50 years. Similarly, McGrattan and Prescott (2003, 2004) find that a decrease in the tax rate on corporate income – one form of capital income – is instrumental in explaining the increase in the value of corporate capital relative to GDP. Overall, we find that the quantitative evidence on the time path of the relevant tax rates to be difficult to ascertain. More work is needed, with an emphasis on matching model and data. The next section considers the effects of endogenous government spending and transitional effects. 2.3.2. Productive government spending A simple balanced growth result. In this section we study a simple one sector model that provides a role for productive government spending. Our discussion follows the ideas in Barro (1990). Assume that firm i’s technology is given by η
1−α−η
yit Akitα hit Gt
,
where kit and hit are the amounts of physical and human capital used by the firm, and Gt is a measure of productive public goods that firms take as given. The government budget constraint is balanced in every period, and it satisfies Gt = τ k rt Kt + τ h wt Ht , where τ k and τ h are the tax rates on capital and income, and rt and wt are rental prices. For simplicity we assume that the instantaneous utility function is given by c1−θ − 1 . 1−θ We also assume that the technologies to produce market goods and human capital are identical. In this case, it is immediate to show that the equilibrium is fully described by (1−α−η)/(α+η) δh − δk = A1/(α+η) ατ k + ητ h × η 1 − τ h κ α/(α+η) − α 1 − τ k κ −η/(α+η) , (1−α−η)/(α+η) −η/(α+η) , κ γ θ = β 1 − δk + α 1 − τ k A1/(α+η) ατ k + ητ h u(c) =
where κ is the physical capital–human capital ratio. Some tedious algebra shows that the growth rate is not a monotonic function of the tax rates. In general, there is no growth when taxes are either too low (not enough public goods are provided) or too high (the private returns to capital accumulation are too low). For intermediate values of the tax rates, growth is positive (if A is sufficiently
26
L.E. Jones and R.E. Manuelli
high). Thus, in general, increases in tax rates need not result in lower growth if they are accompanied by changes in government spending. Thus, a variant of the model with endogenous government spending (or endogenous taxation and optimally chosen government spending) has potential to reconcile positive growth effects associated with the removal of distortions with the U.S. evidence. What does the U.S. evidence show? In the U.S. there is a substantial increase in the ratio of government spending to GDP in the post WWII period on the order of 15%. Even ignoring defense related expenditures, the size of the federal government relative to output is close to 5% in the pre WWII period, and it increases steadily in the post war to reach about 20% of income. Of course, not all forms of government spending are productive, but if the trend in the productive component follows the trend in overall spending, ignoring changes in government spending result in biased estimates of the effects of distortions. The Barro model is silent about the reasons why the desired ratio of (productive) government spending to GDP would increase. For this, it is necessary to have a model of the collective decision making mechanisms which is clearly beyond the scope of this chapter. Progressive taxes and transition effects. Our discussion of the assumptions that suffice for sustained growth clearly shows that homogeneity of degree one is not necessary. In both theoretical and applied work it is common to appeal to linearity in order to ignore transitional dynamics [see Bond, Ping and Yip (1996) and Ladron de Guevara, Ortigueira and Santos (1997) for analysis of the dynamics of endogenous growth models]. However, when taking the model to the data, the assumption that the economy is on the balanced growth path may not be appropriate. In this section we describe the results of Li and Sarte (2001). The basic insight from their model that is relevant for our discussion is that in the presence of heterogeneity in individual preferences and nonlinearities in the tax code, shocks to the tax regime (they consider an increase in the degree of progressivity of the tax code) that ultimately result in a decrease in the growth rate can have basically no effects for several decades. The basic model that they consider is one in which goods are produced according to the following technology Yt AKtα Lt1−α Gt1−α , where Kt is capital at time t, Lt is the flow of labor, and Gt is a measure of productive public goods. All individuals have isoelastic preferences given by u(c) = (c1−θ − 1)/ (1 − θ ), but they differ in their discount factors, βi . Li and Sarte assume that each type has mass 1/N, where N is population. The tax code is nonlinear. Given aggregate income Y , and individual income yi , the tax rate is given by a function τ (z), where z is the ratio of individual to average income. In this application, Lin and Sarte assume that φ yi yi . τ =ζ Y Y
Ch. 1: Neoclassical Models of Endogenous Growth
27
Note that the case of proportional taxes – the case discussed so far – corresponds to φ = 0. In this setting, higher values of φ are interpreted as corresponding to more progressive tax codes. Individual income is defined as the sum of capital and labor income. Government spending is financed with revenue from income taxes. Li and Sarte show that the equilibrium is the solution to the following system of equations:
φ (1−α)/α yi G γ θ = βi 1 − δ + 1 − (1 + φ)ζ αA1/α , i = 1, 2, . . . , I, Y Y 1+φ I G 1 yi = , ζ Y Y N i=1
1=
I i=1
yi 1 . Y N
In this model, changes in the progressivity of the tax code affect the rate of return – this is the standard effect – as well as the distribution of income. It is this last effect that generates the slow adjustment. It is possible to show that an increase in φ decreases long run growth, γ . Li and Sarte explore the dynamic effects of a one time increase in φ that result in a decrease in the growth rate of 1.5%. On impact, output growth increases because since the distribution of income does not adjust immediately, government revenues increase and this, in turn, increases output. As the low discount factor individuals adjust their relative income (an increase in progressivity affects them more than proportionally), government revenue and spending decrease. For parameter values that are designed to mimic the U.S. economy, Li and Sarte find that the half-life of the adjustment is over 40 years. Thus, any test for breaks in the growth rate as suggested by models in which convergence is immediate would conclude (incorrectly) that the tax increase has no effects on growth. It is difficult to evaluate how appropriate the Li and Sarte model is to study the impact of tax reform in the U.S. economy. However, it casts doubt on the approach by Stokey and Rebelo which ignores transitional dynamics. Models that rely on changes in tax rates that, in turn, affect the distribution of income, are consistent with the view that the effects of those changes are not monotone, and that the full impact may not be felt for decades. 2.4. Innovation in the neoclassical model One of the things that seems unsatisfactory to many economists in the presentation up to this point is the starkness with which the technological side of the model is described. As we argued above, the key in improving over the Solow model is to explicitly consider decisions made by private agents about investments they make that cause technology to improve. This both endogeneizes the growth process envisaged by Solow and breaks
28
L.E. Jones and R.E. Manuelli
away from another key assumption of the exogenous growth literature, that technological change happens without any resource cost. But, much of the detail that one thinks about as being an important part of the innovation process seems to be missing from the simple convex models of growth described above. The idea that innovation is carried out by specialized researchers who pass on their newfound knowledge to production line workers is just one example of this. Indeed, one question is whether or not that kind of structure is consistent with convex models of growth at all. Because of this, in this section we describe a variant of the models presented in the last section that is more directed at identifying innovation as a special activity. The purpose of this exercise is not to fully exhaust the possibilities, but rather to show the reader that more is possible with the class of convex models than one might first think. In particular, since the model we will analyze is convex, standard price taking behavior is consistent with equilibrium behavior. In this sense, the example we will present is similar to the ideas developed by Boldrin and Levine (2002). There are many models of innovation that do not have convex technology sets [e.g., see the surveys in Barro and Sala-i-Martin (1995) and Aghion and Howitt (1998)]. In this setting, standard price taking behavior is either not consistent with equilibrium in those settings or they must include external effects. Because of this, all policy experiments in those models mix two conceptually distinct aspects of policy, the desire to correct for monopoly power and/or external effects and the distortionary effects of ‘wedges’ (e.g., taxes). This, in turn, implies that the answers to questions about the effects of alternative policies on both the incentives to innovate and overall welfare depends on the details of the specifications of external effects (e.g., do other researchers learn new innovations for free after one month, or one decade) and/or market power (e.g., is there only one researcher at the frontier and so a monopoly analysis is in order, or are there two, or many). Thus, one thing that a convex model of innovation has to add is answers to some of these questions which are less dependent on those details. 2.4.1. Notation We will follow the notation above as closely as is possible. We assume that there are two types of labor supply available, researchers and workers. Each individual of each type has his own level of knowledge. We will assume that there is a continuum of identical households each with some researcher time and some worker time to supply to the market. These are given by L1 researcher hours per household, and L2 worker hours per household, where L = L1 + L2 is total labor supply within the household. We will assume that L1 and L2 are fixed, with no ability to move hours between them. (In this sense, it might be easier to think of the household as being made up of L1 researchers and L2 workers.) Each household has the level of knowledge Ht that they can use with researcher hours during period t. Thus, if households are symmetric, Ht symbolizes the absolute frontier of what ‘society’ knows at date t. Similarly, the level of knowledge for the average
Ch. 1: Neoclassical Models of Endogenous Growth
29
worker hour at date t is denoted by ht . This will represent the average knowledge of those workers that work in the final goods sector below. Final consumption at date t is denoted by ct . We abstract from physical capital to simplify. Production functions. We will assume that: Ht+1 = Ht + AH LH 1t Ht , α 1−α , ht+1 = (1 − δh )ht + Ah Lh1t Ht Lh2t ht ct = Ac Lc2t ht , h LH 1t + L1t = L1 ,
Lh2t + Lc2t = L2 . This formulation is equivalent to one in which quality adjusted labor of the form Z = LH (resp. Z = Lh) is employed in each activity. The idea here is that IH = AH LH 1t Ht is new research and development or innovation, increasing H corresponds to learning more at society’s highest level. Note that we have assumed that there is no depreciation – the level of frontier knowledge does not go backward (positive depreciation is easily included). If no innovation is done, LH 1t = 0 for all t, Ht+1 = Ht for all t, that is, the frontier is static. In this case, ht would also be bounded, and hence the level of output would be bounded above. In this sense, innovation is necessary for growth to occur in this model. Similarly, we think of the Ih = Ah (Lh1t Ht )α (Lh2t ht )1−α technology as Education and/or Worker Training. This is where family members at the frontier spend part of their time educating the workers from the family on the use of new techniques. The more time researchers spend in Ih , the less time they have to spend in IH , and hence workers are better prepared and more productive, but the frontier moves out more slowly. Note that increasing Lh2t ht , holding Lh1t Ht constant increases total output of worker productive knowledge (new h) but lowers the average product of frontier knowledge workers in educating (bigger classes give more total new training, but less output per student). Symmetrically, the more time that production line workers spend learning new knowledge, the less time they have available for production of current consumption goods. Preferences. We assume that each researcher/worker supplies one unit of labor to the market inelastically and that each has preferences of the form: β t u(ct ), U (c) = t
where u(c) = c1−σ /(1 − σ ). This model, although different in detail, shares one common critical feature with those above: linearity in the reproducible factors.
30
L.E. Jones and R.E. Manuelli
2.4.2. Balanced growth properties of the model Like many Ak, style models, this one has the feature that it converges to a Balanced Growth Path. Indeed if the initial levels of relative human capitals (H0 / h0 ) are right the economy is on the BGP in every period. Standard techniques can be used to characterize this BGP. After some algebra, we find: γ σ = β 1 + AH LH 1 . This equation gives γ as a function of the basic parameters σ , β, AH , and L1 . By construction, the comparative statics of growth rates with respect to the deep parameters of the model are identical to what one would find in an Ak model. The one difference is the inclusion of the endowment of researcher hours, L1 , note that γ is increasing in L1 . That is, if one country had a higher proportion of researchers in its population, output would grow faster. Since γ does not depend on the other parameters of the model, it can be shown that the only way income taxes affect growth rates here is through their effect on the R&D sector. That is, if we have a linear income tax (at rate τ ) either on income generated in all sectors, or on income generated only in the H sector the growth rate will fall to: γτσ = β 1 + (1 − τ )AH LH 1 . In particular, if income from the h and/or c sectors are taxed, but that from the H sector is not, there are no effects on growth. This is reminiscent of the Lucas (1988) model and the 2-sector model in Rebelo (1991). The amount of time spent on R&D on the BGP is given by: LH 1 = =
[(β[1 + A∗H ])1/σ − 1] [(β[1 + AH L1 ])1/σ − 1] L = L1 1 A∗H AH L1 [(β[1 + AH L1 ])1/σ − 1] . AH
Thus, if we compare two countries with different discount factors, but identical in other respects, the one with the higher β will devote a higher fraction of its researcher time to innovation and a lower percentage to teaching. This causes worker productivity in the consumption sector to be lower at first (and consumption as well), but growing faster and hence, eventually overtaking the low β country. As a second point, note that increases in Ah do not change γ (and neither do changes in Ac ). Thus, in this case, LH 1 is not affected and so the time series of Ht will be identical. This implies that ht must be higher. Thus, wages of both researchers and workers will be higher. This is similar in spirit to the result in Boldrin and Levine (2002) that improvements in the copying technology raises the value of being an innovator. Since the only ‘copying’ being done here is the passing on of new knowledge to final goods workers, this is analogous in this setting.
Ch. 1: Neoclassical Models of Endogenous Growth
31
These are simple comparative statics exercises which are meant only to show that much intuition about the process of innovation, and its comparative statics properties with respect to incentives can be illustrated in this class of models. There are many interesting extensions of the analysis that one could imagine. These include heterogeneity among households (e.g., some researcher households, some worker households), the inclusion of uncertainty about the results of researcher time (and the questions that this raises about ex post hold up problems when one researcher is the ‘first’ discoverer), the training of researchers by other researchers when they have different Ht ’s, the inclusion of more than one good or process, what types of Industrial Organization are possible through decentralizations of the allocation as a competitive equilibrium (e.g., firms specializing in R&D vs. each firm having an R&D division), etc. But, the reader can see that much of the analysis will go through unchanged. Notable exceptions are when there are assumed to be external effects in the learning process. The simplest example of this here would be to assume that h = H no matter what Lh2 is. In this case, unless this is completely internalized within a firm (i.e., there are no spillovers across firms) the Planner’s problem will not be implementable as a competitive equilibrium. 2.4.3. Adding a non-convexity Most models of innovation differ from the one outlined above in that they assume that there is a non-convexity in the innovation technology. There are two ways to include this in the specification above, and the differences between them highlight a key question about innovation. These are: H ∗ Ht + AH LH 1t Ht if L1t L , Ht+1 = (8) ∗ if LH Ht 1t < L and
Ht+1 =
∗ Ht + AH (LH 1t − L )Ht
∗ if LH 1t L ,
Ht
∗ if LH 1t < L .
(9)
Although these two specifications look similar, they differ in one key aspect. The ∗ technology in (8) is convex anytime R&D is ‘active’ (i.e., LH 1t L ). The technology in (9) has constant marginal costs when R&D is active, but features a set-up cost as well (given by L∗ denominated in labor units). Technology (9) is the specification that is most commonly employed in the R&D literature while that in (8) is similar in spirit to that used in Boldrin and Levine. Because of this difference, all of the analysis outlined above can be used if the technology is that given in (8) so long as in the solution to the ∗ planner’s problem we have that N LH 1t L for all t where N is the number of households. That is, the allocation can be supported as a competitive equilibrium with price
32
L.E. Jones and R.E. Manuelli
taking behavior, etc. There are some restrictions on the implicit Industrial Organization ∗ in the equilibrium however. For example, if N LH 1t = L for all t it follows that there can be at most one R&D firm in any equilibrium (or one firm with an R&D division). This, were it true, would cause serious concern for the price-taking assumption in the decentralization. One interesting implication of this model is that whether or not the solution to the planning problem above (without the non-convexity) describes the competitive allocation depends on the size of the country, N . Thus, large countries would, in equilibrium, conduct R&D while small countries would not. Adding in a fourth sector in which researchers in large countries could train researchers in small countries would be a natural extension in which R&D was done in large countries, these researchers train high H workers from small countries (e.g., in Engineering schools), those newly trained ‘researchers’ return to their home countries where they subsequently train production line workers, etc. This description of equilibrium cannot be true for (9), however. Price taking behavior in this setting implies that prices for the rental of new knowledge equal their marginal cost of production. This implies that there is no way to recover the set up cost of researchers spending L∗ hours. Thus, there can be no competitive equilibrium. It follows that there must be some monopoly rent generated somewhere to decentralize any allocation. Typically this will be accompanied by inefficiencies and incorrect incentives to conduct R&D.
3. Fluctuations and growth 3.1. Introduction In this section we describe existing results on the effects of ‘volatility’, both in technologies and policies, on the long-run growth rate. We start with a brief summary of the empirical research in this area, and we then describe some simple theoretical models that are useful in understanding the empirical results. We end with the description of some recent work based on the theoretical models but aimed at evaluating their ability to quantitatively match the growth observations. As before, we ignore models based on aggregate non-convexities, and with non-competitive market structures. 3.2. Empirical evidence A relatively small (but growing) empirical literature has tried to shed light on the relationship between ‘instability’ and growth. This literature has concentrated on estimating reduced form models that try to capture, with varying degrees of sophistication, how ‘volatility’ (defined in a variety of different ways) affects long-run growth. Kormendi and Meguire (1985) is probably the earliest study in this literature. They consider a sample of 47 countries with data covering the 1950–1977 period. Their
Ch. 1: Neoclassical Models of Endogenous Growth
33
methodology is to run a cross-country growth regression with the average (over the sample period) growth rate as the dependent variable, and a number of control variables, including the standard deviation of the growth rate (one measure of instability), as well as the standard deviations of policy variables such as the inflation rate and the money supply. Kormendi and Meguire find that the coefficient of the volatility measure (the standard deviation of the growth rate) is positive and significant. Thus, a simple interpretation of their results is that more volatile countries – as measured by the standard deviation of their growth rates – grow at a higher rate. Grier and Tullock (1989) use panel data techniques on a sample of 113 countries covering a period from 1951 to 1980. Their findings on the effect of volatility on growth are in line with those of Kormendi and Meguire. They find that the standard deviation of the growth rate is positively, and significantly, associated with mean growth rates. Ramey and Ramey (1995) first report the results of regressing mean growth on its standard deviation on a sample of 92 countries as well as a subsample of 25 OECD countries, covering (approximately) the 1950–1985 period. They find that for the full sample the estimated effect of volatility is negative and significant, while for the OECD subsample the point estimate is positive, but insignificant. In order to allow for the variance of the innovations to the growth rate to be jointly estimated with the effects of volatility, Ramey and Ramey posit the following statistical model γit = βXit + λσi + uit ,
(10)
where Xit is a vector of variables that affect the growth rate and uit = σi it ,
it ∼ N (0, 1).
(11)
The model is estimated using maximum likelihood. The control variables used were the (average) investment share of GDP (Average I /Y ), average population growth rate (Average γPop ), initial human capital (measured as secondary enrollment rate, H0 ), and the initial level of per capita GDP (Y0 ). They study separately the full sample (consisting of 92 countries) as well as a subsample of 25 OECD (more developed) economies. Their results are reproduced in columns (1) and (3) of Table 1. For both sets of countries, Ramey and Ramey find that the standard deviation of the growth rate is negatively related to the average growth rate. However, for the OECD subsample, the coefficient is less precisely estimated (even though the point estimate is larger in absolute value). Ramey and Ramey also consider more ‘flexible’ specifications that try to capture differences across countries in the appropriate forecasting equations. Considering the most parsimonious version of their model, the estimated effect of volatility on growth is still positive. However, the strength of the estimated relationship is reversed: for the OECD subsample the point estimate is four times the size of the estimate for the full sample and highly significant. In more recent work, Barlevy (2002) reestimates the Ramey and Ramey model with one change: he adds the standard deviation of the logarithm of the investment-output ratio (σln(I /Y ) ) as one of the explanatory variables. Barlevy hypothesizes that this variable can capture non-linearities in the investment function. His results, using the same
34
L.E. Jones and R.E. Manuelli Table 1 Growth and volatility I
Variables
(1) (92-Country) N = 2,184
(2) (92-Country) N = 2,184
(3) (OECD) N = 888
(4) (OECD) N = 888
Constant
0.07 (3.72) −0.21 (−2.61) 0.13 (7.63) −0.06 (−0.38) 0.0008 (1.18) −0.009 (−3.61)
0.08 (3.73) −0.109 (−1.22) 0.12 (6.99) −0.115 (−0.755) 0.0007 (1.03) −0.009 (−3.53)
0.16 (5.73) −0.39 (−1.92) 0.07 (2.76) 0.21 (0.70) 0.0001 (2.00) −0.017 (−5.70)
0.16 (4.48) −0.401 (−1.93) 0.071 (2.67) 0.230 (0.748) 0.0001 (1.954) −0.017 (−4.7445)
–
−0.023 (1.81)
σi Average I /Y Average γPop H0 Y0 σln(I /Y )
0.007 (0.22)
Note: t-statistics in parentheses. Source: Columns (1) and (3) – Ramey and Ramey (1995), columns (2) and (4) – Barlevy (2002).
basic data as Ramey and Ramey are in columns (2) and (4) of Table 1.6 For the full 92-country sample, the introduction of this measure of investment volatility halves the size of the coefficient of σi , and it is no longer significant at conventional levels. The coefficient on σln(I /Y ) is negative and significant (at 5%). For the OECD sample, the addition of σln(I /Y ) does not affect much the estimate of the effect of σi on growth. However, Barlevy points out that this is finding is not robust, since eliminating two outliers, Greece and Japan where high volatility of the investment share seems to be due to transitional dynamics, implies that neither the volatility of the growth rate nor σln(I /Y ) are significant.7 One possible explanation for the differences in the estimates of the effects of volatility on growth found in Kormendi and Meguire, Grier and Tullock and Ramey and Ramey, is – as pointed out by Ramey and Ramey and Barlevi – that Kormendi and Meguire and Grier and Tullock include among their explanatory variables the standard deviations of policy variables that could be proxying for σln(I /Y ) . Kroft and Lloyd-Ellis (2002) also start from the basic statistical model of Ramey and Ramey but offer a different way of decomposing volatility. They hypothesize that uncertainty can be split into two orthogonal components: uncertainty about changes in 6 We thank Gadi Barlevy for providing us the estimated coefficient for the control variables. 7 The point estimates are negative but insignificant.
Ch. 1: Neoclassical Models of Endogenous Growth
35
Table 2 Growth and volatility II Independent variable
92-Country sample (2,208 observations)
OECD sample (888 observations)
0.00132 (0.022) 2.63 (4.69) −2.65 (−6.35) −0.01 (−0.26) 0.58 (1.24) 0.001 (0.66) 0.002 (0.25)
0.095 (1.89) 0.90 (1.44) −1.11 (−2.33) −0.004 (−0.073) 0.28 (0.62) −0.00001 (−0.096) −0.0008 (−1.30)
Constant Within-phase volatility (σiw ) Between-phase volatility (σib ) Average investment share of GDP Average population growth rate Initial human capital Initial per capita GDP
Note: t-statistics in parentheses. Source: Kroft and Lloyd-Ellis (2002).
regime (e.g., expansion–contraction) and fluctuations within a given regime. To this end, they generalize the empirical specification of the Ramey and Ramey statistical model to account for this. They assume that γist = βXit + λw σiw + λb σib + υist ,
(12a)
υist = σiw it + µis , it ∼ N (0, 1),
Tie µis = µie with probability pi = T , µir with probability 1 − pi .
(12b) (12c)
Kroft and Lloyd-Ellis interpret the standard deviation of the random variable µis , σib – which they assumed observed by the economic agents but unobserved by the econometrician – as a measure of variability between regimes, while σiw is viewed as the within-regime variability. Kroft and Lloyd-Ellis estimate their model by maximum likelihood using the same sample as Ramey and Ramey. The results are in Table 2. The major finding is that the ‘source’ of volatility matters: Increases in σiw – the within phase standard deviation – have a positive impact on growth for the full sample. For the OECD, the coefficient estimate is still positive but about one third of the size. The effect of the between-phase volatility, σib , is negative in both cases. However, the effects are stronger for the full sample. It is not easy to interpret the phases identified by Kroft and Lloyd-Ellis in terms of a switching model because their estimation procedure assumes that the econometrician can identify whether a particular period corresponds
36
L.E. Jones and R.E. Manuelli
to either a recession or an expansion.8 Kroft and Lloyd-Ellis also use the same controls as Ramey and Ramey. However, they find that, when the two variances are allowed to differ, none of the control variables is significant. It is not clear why this is the case. One possibility is that the ‘phases’ that they identify are correlated with the control variables (this seems likely situation in the case of investment). Another possibility is that the control variables, in the Ramey and Ramey formulation, capture the nonlinearity associated with the regime shift and that, once the shifts are taken into account, the control variables have no explanatory power. In any case, this illustrates a point that we will come back to: the fragility of the “growth” regressions suggest that better theoretical models are necessary to more provide restrictions that will allow to identify the parameters of interest. The results of Ramey and Ramey and Kroft and Lloyd-Ellis are consistent with the existence of nonlinearities in the relationship between measures of instability and growth. Fatás (2001) estimates a number of different specifications of the relationship between instability and growth. His approach is to run standard cross country regressions. His data set is taken from the most recent version of the Heston–Summers sample and includes 98 countries with information covering the period 1950–1998. His estimates (see Table 3) support the view that the effect of volatility on growth is nonlinear. Using Fatás’ basic estimate – shown in column (1) of Table 3 – the pure effect of volatility is negative with a coefficient of −2.772 indicating that a one standard deviation increase in volatility reduces the growth rate by over 2.5%. However, the interaction term, corresponding to the variable Volatility · GDP is positive and equal to 0.340. According to these estimates, the net effect of σi on γi for the richest countries in the data is positive and greater than 0.3. For the less developed countries the estimate of the effect of volatility is negative. Columns (2) and (3) use other measures of non-linearity (initial per capita GDP and M3/Y , a measure of financial development), with similar outcomes: In all cases there is a significant effect, and increases in volatility are less detrimental to growth – and could even have a positive effect – the more developed a country is according to the proxy variables. Martin and Rogers (2000) also study the relationship between the standard deviation of the growth rate and its mean, in a cross section of countries and regions. They study two samples – European regions and industrialized countries – and in both cases they find a negative relationship between σγ and γ . However, when they consider a sample of developing countries the point estimates are positive, but in general insignificant. It is not easy to explain the differences between Ramey and Ramey, Fatás and Martin and Rogers. The period used to compute the growth rates (1962–1985 for Ramey and Ramey, 1950–1998 for Fatás and 1960 to 1988 for Martin and Rogers), and the
8 Kraft and Lloyd-Ellis estimate the probabilities p as the fraction of the time that an economy spends in i the recession “phase”, defined as periods of negative output growth. Thus, not only is the process assumed to be i.i.d. but the transition probabilities are not jointly estimated with the parameters.
Ch. 1: Neoclassical Models of Endogenous Growth
37
Table 3 Growth and volatility III Independent variable Volatility (σi ) GDP per capita (1960) Human capital (1960) Average investment share of GDP Average population growth rate Volatility · GDP Volatility · GDP (1960) Volatility · M3/Y R2
(1)
(2)
(3)
−2.772 (0.282) −2.229 (0.235) 0.037 (0.015) 0.083 (0.013) −0.624 (0.153) 0.340 (0.036) –
−1.700 (0.645) −1.856 (0.422) 0.040 (0.018) 0.143 (0.021) −0.562 (0.205) –
−0.270 (0.091) −0.953 (0.220) 0.026 (0.017) 0.120 (0.024) −0.465 (0.465) – –
–
0.212 (0.082) –
0.77
0.58
0.004 (0.001) 0.57
Note: Sample 1950–1998. Robust standard errors in parentheses. Source: Fatás (2001).
set of less developed countries included (68 in Ramey and Ramey’s study, and 72 in Martin and Rogers’) are fairly similar. The two studies differ on their definition of the growth rate (simple averages in the Ramey and Ramey and Fatás papers, and estimated exponential trend in Martin and Rogers), and in the variables that are used as controls. However, it is somewhat disturbing that what appear, in the absence of a theory, as ex-ante minor differences in definitions can result in substantial differences in the estimates. Siegler (2001) studies the connection between volatility in inflation and growth rates and mean growth for the pre 1929 period. Specifically, he uses panel data methods for a sample of 12 (presently developed) countries over the 1870–1929 period. He finds that volatility and growth are negatively correlated, and this finding is robust to the inclusion of standard growth regression type of controls. Dawson and Stephenson (1997) estimate a model similar to (10) and (11) applied to U.S. states. They use the average (over the 1970–1988 period) growth rate of gross state product per worker for U.S. states as their growth variable, and its standard deviation as a measure of volatility. In addition, they include in their cross-sectional regression the standard (in growth regressions) control variables (investment rate, initial level of gross state product per worker, labor force growth rate, and initial human capital). Dawson and Stephenson find that volatility has no impact on the growth rate, once the other effects are included. Unfortunately, they do not report the ‘raw’ correlation between
38
L.E. Jones and R.E. Manuelli
mean growth and its standard deviation. Thus, it is not possible to determine if the lack of significance is due to the use of controls, or is a more robust feature of U.S. states growth performance. Mendoza (1997) differs from the previous studies in terms of his definition of instability. Instead of the standard deviation of the growth rate, which, in general, is endogenous, he identifies instability with the standard deviation of a country’s terms of trade. He estimates a linear model using a cross section of countries and finds a negative relationship between instability and growth. His sample is limited to only 40 developed and developing countries, and it only covers the period 1971–1991. A fair summary of the existing results is that there is no sharp characterization of the relationship between fluctuations and growth. Variation across studies in samples or specifications yield fairly different results. Moreover, the findings do not seem robust to details of how the statistical model is specified. Are the empirical findings of the channel through which uncertainty affects growth more robust? Unfortunately, the answer is negative. Ramey and Ramey find that volatility – measured as the standard deviation of the growth rate – does not affect the investment–output ratio. More recently, Aizenman and Marion (1999) find that volatility is negatively correlated with investment, when investment is disaggregated between public and private. Fatás estimates a non-linear model of the effect of volatility on investment. He finds that increases in volatility decrease investment in poor countries, but that the opposite is true in high income countries. Thus his findings are consistent with the view that changes in volatility affect mean growth rates through (at least partially) their impact upon investment decisions. How should these empirical results be interpreted? Even though it is tempting to take one’s preferred point estimate as a measure of the impact of fluctuations (or business cycles) on growth there are two problems with this approach. First, the empirical estimates are not robust to the choice of specification of the reduced form. Second, and more important in our opinion, is that from the point of view of policy design, the relevant measures of volatility is the – in general unobserved – volatility in policies and technologies. In most models, the growth rate (and its standard deviation) are endogenous variables and, as usual, the point estimate of one endogenous variable on another is at best difficult to interpret. One way of contributing to the interpretation of the empirical results is to study what simple theoretical models predict for the estimated relationships. In the next section we present a number of very simple models to illustrate the possible effects of volatility in fundamentals on mean growth. In the process, we find that it is very difficult to interpret the empirical findings. To put it simply, there are theoretical models that – depending on the sample – do not restrict the sign of the estimated coefficient of the standard deviation of the growth rate on its mean. Moreover, the sign and the magnitude of the coefficient is completely uninformative to determine the effect of volatility on growth.
Ch. 1: Neoclassical Models of Endogenous Growth
39
3.3. Theoretical models The analysis of the effect of uncertainty on growth can be traced to the early work of Phelps (1962), and Levhari and Srinivasan (1969) who studied versions of the stochastic consumption-saving problem that are similar to the linear technology versions of endogenous growth models. More recently, Leland (1974), studies a stochastic Ak model, and he shows that the impact of increased uncertainty on the consumption/output ratio depends on the size of the coefficient of risk aversion. Even in deterministic versions of models that allow for the possibility of endogenous growth, existence of equilibria (and even optimal allocations) requires strong assumptions on the fundamentals of the economy [see Jones and Manuelli (1990) for a discussion]. At this point, there are no general results on existence of equilibrium in stochastic versions of those models. In special cases, most authors provide conditions under which an equilibrium exists [see Levhari and Srinivasan (1969), Mendoza (1997), Jones, Manuelli and Siu (2003), Jones et al. (2003) for various versions]. A recent, more general result is contained in de Hek and Roy (2001). These authors consider fairly general utility and production functions, but limit themselves to i.i.d. shocks. It is clear that more work is needed. In what follows, we will describe a general linear model and we will use it to illustrate the predictions of the theory for the relationship between mean growth rates and their variability. To simplify the presentation we switch to a continuous time setting. In order to obtain closed-form solutions we specialize the model in terms of specifying preferences and technology. Moreover, we will limit ourselves to i.i.d. shocks. Generalizations of these assumptions are discussed in the section that presents quantitative results. 3.4. A simple linear endogenous growth model We begin by presenting a stochastic analog of a standard Ak model with a ‘twist’. Specifically, we consider the case in which there are multiple linear technologies, all producing the same good. In order to obtain closed-form results we specify that the utility of the representative household is given by ∞ 1−θ −ρt ct U =E (13) dt F0 . e 1−θ 0 We assume that each economy has two types of technologies to produce consumption (alternatively, the model can be interpreted as a two sector model with goods that are perfect substitutes). Output for each technology satisfies dkt = (A − δk )kt − c1t dt + σk kt dWt + ηk kt dZtk , (14a) b dbt = (r − δb )kt − c2t dt + σb bt dWt + ηb bt dZt , (14b) where (Wt , Ztk , Ztb ) is a vector of three independent standard Brownian motion processes, and kt and bt are two different stocks of capital. This specification assumes
40
L.E. Jones and R.E. Manuelli
that each sector is subject to an aggregate shock, Wt , as well as sector (or technology) j specific shocks, Zt . To simplify the algebra, we assume that capital can be costlessly reallocated across technologies, and we denote total capital by xt ≡ kt + bt . Setting (without loss of generality) kt = αt xt (and, consequently bt = (1 − αt )xt ) it follows that total capital evolves according to dxt =
αt (A − δk ) + (1 − αt )(r − δb ) xt − ct dt + αt σk + (1 − αt )σb dWt + αt ηk dZtk + (1 − αt )ηb dZtb xt .
(15)
Given the equivalence between equilibrium and optimal allocations in this convex economy, we study the solution to the problem faced by a planner who maximizes the utility of the representative agent subject to the feasibility constraint. Formally, the planner solves max E 0
∞
e−ρt
ct1−θ dt F0 , 1−θ
subject to (15). Let the value of this problem be V (x). Then, it is standard to show that the solution to the planner’s problem satisfies the Hamilton–Jacobi–Bellman equation 1−θ V (x)x 2 2 c + V (x) µ(α)x − c + σ (α) , ρV (x) = max c,α 1 − θ 2 where µ(α) = r + α(A − r) − αδk + (1 − α)δb , 2 σ 2 (α) = ασk + (1 − α)σb + α 2 ηk2 + (1 − α)2 ηb2 .
(16a) (16b) 1−θ
It can be verified that the solution is given by V (x) = v x1−θ , where
ρ − (1 − θ )[µ(α ∗ ) − δ(α ∗ ) − θ σ v= θ
2 (α ∗ )
2
]
−θ (17)
and δ(α) = αδk + (1 − α)δb . The optimal decision rules are α∗ =
A−δk −(r−δb ) θ
− σb (σk − σb ) + ηb2
(σk − σb )2 + ηb2 + ηk2
ρ − (1 − θ )[µ(α ∗ ) − δ(α ∗ ) − θ σ c= θ
,
2 (α ∗ )
2
(18a) ]
x.
(18b)
Ch. 1: Neoclassical Models of Endogenous Growth
41
For the solution to be well defined it is necessary that ρ − (1 − θ )[µ(α ∗ ) − δ(α ∗ ) − 2 ∗) θ σ (α 2 ] > 0, which we assume. (In each case we make enough assumptions to guarantee that this holds.)9 It follows that the equilibrium stochastic differential equation satisfied by aggregate wealth is given by µ(α ∗ ) − (δ(α ∗ ) + ρ) σ 2 (α ∗ ) xt dt dxt = − (1 − θ ) θ 2 + α ∗ (σk − σb ) + σb dWt + α ∗ ηk dZtk + (1 − α ∗ )ηb dZtb xt , (19) and the instantaneous growth rate of the economy, γ , and its variance, σγ2 , satisfy σ 2 (α ∗ ) µ(α ∗ ) − (δ(α ∗ ) + ρ) − (1 − θ ) , θ 2 2 σγ2 = α ∗ (σk − σb ) + σb + α ∗2 ηk2 + (1 − α ∗ )2 ηb2 .
γ =
(20a) (20b)
One is tempted to interpret (19) as the theoretical analog of (10) by defining the stochastic growth rate as γt =
dxt . xt
Given this definition, the discrete time – with period length equal to one – version of the stochastic process followed by the growth rate is σγ2 µ(α ∗ ) − (δ(α ∗ ) + ρ) − (1 − θ ) + εt , θ 2 ∗ εt = α (σk − σb ) + σb dWt + α ∗ ηk dZtk + (1 − α ∗ )ηb dZtb .
γt =
(21a) (21b)
This simple model driven by i.i.d. shocks has a stark implication: the growth rate is i.i.d. and is independent of other endogenous (or exogenous) variables, except through the joint dependence on the error term. Using panel data, it is relatively easy to reject this implication. This, however, is not an intrinsic weakness of this class of models. The theoretical setting can be generalized to include serially correlated shocks and a nonlinear structure, which could account for “convergence” effects, and would provide a role for lagged dependent variables. However, generalizing the theoretical model comes at the cost of not being able to discuss the impact of different factors on the growth rate, except numerically.
9 In endogenous growth models existence of an equilibrium is not always guaranteed. The main problem is that with unbounded instantaneous utility and production sets, utility can be infinite. For a discussion of some conditions that guarantee existence see Jones and Manuelli (1990) and Alvarez and Stokey (1998). The key issue is that the return function is unbounded above when 0 < θ < 1, and unbounded below if θ > 1. In this setting, it can be shown that c > 0 is equivalent to ensuring boundedness.
42
L.E. Jones and R.E. Manuelli
What is the (simple) class of model that we study useful for? We view the class of theoretical models that we present as more appropriate to discuss the implications of the ∗ ∗ )+ρ) − theory for cross section regressions since, in this case, the constant µ(α )−(δ(α θ σ2
(1 − θ ) 2γ can be correlated with other variables like the investment–output ratio. Even though there is a formal similarity between (21) and (10)–(11), the theoretical model suggests that the simple approach that ignores that the same factors that affect σγ , also influence the true value of β in (10) can result in incorrect inference. Alternatively, the “deep parameters” are not the means and the standard deviation of the growth rates. They are the means and standard deviations of the driving stochastic processes. In terms of those parameters, the “true” model is non-linear. Whether the model in (21) implies a positive or negative relationship between fluctuations and growth depends on the sources of shocks. At this general level it is difficult to illustrate this point, but we will come back to it in the context of specific examples. It is not obvious how to define the investment ratio in this model. The change in cumulative investment in k, Xk , is given by dXkt = δk kt dt + dkt , while the change in total output can be defined as10 dYt = µ(α ∗ )xt dt + σγ xt dMt , where Mt is a standard Brownian motion defined so that σγ dMt = α ∗ (σk − σb ) + σb dWt + α ∗ ηk dZtk + (1 − α ∗ )ηb dZtb . In order to avoid technical problems, we consider a discrete time approximation in which the capital stocks change only at the beginning of the period. The investmentoutput ratio (for physical capital) is given by zt =
γ + δk + σγ εt , µ(α ∗ ) + σγ εt
where εt is the same noise that appears in (12). Since the previous expression is nonlinear, we approximate it by a second order Taylor expansion to obtain zt =
σγ2 [µ(α ∗ ) − (γ + δ)] 2 σγ [µ(α ∗ ) − (γ + δk )] γ + δk + ε − εt . t µ(α ∗ ) µ(α ∗ )2 µ(α ∗ )3
The mean investment ratio, which we denote z, is given by σγ2 σγ2 γ + δk − 1 + . z= µ(α ∗ ) µ(α ∗ )2 µ(α ∗ )2
(22)
(23)
10 This is not the only possible way of defining output. It assumes that the economy two sectors (or technolo-
gies). However, another interpretation of this basic framework considers bt as bonds, and kt as the only real stock of capital. We will be precise about the notion of output in each application.
Ch. 1: Neoclassical Models of Endogenous Growth
43
Given this approximation, the model implies that the covariance between the growth rate and the investment–output ratio is cov(γt , zt ) =
σγ2 [µ(α ∗ ) − (γ + δk )]
µ(α ∗ )2 while the standard deviation of zt is
,
(24)
σγ [µ(α ∗ ) − (γ + δk )] (1 + µ(α ∗ )2 )1/2 (25) . µ(α ∗ ) µ(α ∗ )2 Simple algebra shows that, given that the existence condition (17) is satisfied, cov(γt , zt ) > 0. Thus, in a simple regression, the investment ratio has to appear to affect positively growth. At this general level it is more difficult to determine if high σz economies are also high γ economies. The problem is that there are a number of factors that jointly affect γ and σz . In order to be more precise, it is necessary to be specific about the sources of heterogeneity across countries. We will be able to discuss the sign of this relationship in specific contexts. We now use this ‘general’ model to discuss – in a variety of special cases – the connection between the variability of the growth rate of output and its mean. σz =
3.4.1. Case 1: An Ak model Probably the simplest model to illustrate the role played by differences in the variability of the exogenous shocks across countries is the simple Ak model. Even though it is a special case of the model described in the previous section, it is useful to describe the technology in a slightly different way. Let the feasibility constraint for this economy be given by
t
t
t
t ˆ s ds + ˆ s dWs σy Ak cs ds + dXks . Ak 0
0
0
0
The left-hand side of this condition is the accumulated flow of output until time t, and the right-hand side is the accumulated uses of output, consumption and investment. The law of motion of capital is dkt = −δk kt dt + dXkt , where δk is the depreciation rate. Expressing the economy’s feasibility constraint in flow form, and substituting in the law of motion for physical capital, the resource constraint satisfies ˆ t dWt . dkt = Aˆ − δk kt − ct dt + σy Ak (26) The planner’s problem – which coincides with the competitive equilibrium in this economy – is to maximize (13) subject to (26). This problem resembles the more general model we introduced in the previous section if we set ηb = σb = ηk = 0, and A = Aˆ − δk ,
44
L.E. Jones and R.E. Manuelli
ˆ σk = σy A. In addition, we need to make sure that the “b” technology is not used in equilibrium. A simple way of guaranteeing this is to view r − δb as endogenous, and to choose it so that, in equilibrium, α ∗ = 1; that is, all of the investment is in physical capital. It is immediate to verify that this requires r − δb = Aˆ − δk − θ σy2 Aˆ 2 . In this case is it follows that xt = kt and the formulas in (20) imply that the mean growth rate and the variance of the growth rate satisfy σγ2 Aˆ − (ρ + δk ) − (1 − θ ) , θ 2 2 2 ˆ2 σγ = σy A . γ =
This result, first derived by Phelps (1962) and Levhari and Srinivasan (1969), shows that, in general, the sign of the relationship between the variance of the technology shocks, σy2 , and the growth rate is ambiguous: • If preferences display less curvature than the logarithmic utility function, i.e. 0 < θ < 1, increases in σy are associated with decreases in the mean growth rate, γ . • If θ > 1, increases in σy are associated with increases in the mean growth rate, γ . • In the case in which the utility function is the log (this corresponds to θ = 1) there is no connection between fluctuations and growth. The basic reason for the ambiguity of the theoretical result is that the total effect of a change in the variance of the exogenous shocks on the saving rate – and ultimately on the growth rate – can be decomposed in two effects that work in different directions: • An increase in the variance of the technology makes acquiring future consumption less desirable, as the only way to purchase this good is to invest. Thus, an increase in variance of the technology shocks has a substitution effect that increases the demand for current (relative to future) consumption. This translates into a lower saving and growth rates. • On the other hand, an increase in the variability of the exogenous shocks induces also an income effect. Intuitively, for concave utility functions, the fluctuations of the marginal utility decrease with the level of consumption. Thus, the (negative) effect of fluctuations is smaller when consumption is high. This income effect increases savings, as this is the only way to have a ‘high’ level of consumption (i.e. to spend more time on the relatively flat region of the marginal utility function). The formula we derived shows that the relative strength of the substitution and income effects depends on the degree of curvature of the utility function: if preferences have less curvature than the logarithmic function, the substitution effect dominates and increases in the variance of the exogenous shocks reduce growth. If the utility of the representative agent displays more curvature than the logarithmic function, the income effect dominates and the relationship between fluctuations and growth is positive.
Ch. 1: Neoclassical Models of Endogenous Growth
45
In this simple economy, the variance of the technology shock, σy2 , and the variance ˆ 11 If one views the differof the growth rate of output, σγ2 , coincide up to scale factor A. 2 12 ences across countries as due to differences in σy , the theoretical model implies that the true regression equation is very similar to the one estimated in the empirical studies. The only difference is that the theory implies that it is σγ2 , and not σγ , that enters the right hand side of (10). If we use this model to interpret the results of Ramey and Ramey (1995), one must conclude that the negative relationship between mean growth and its standard deviation is evidence that preferences have less curvature than the logarithmic utility, i.e. 0 < θ < 1. On the other hand, the Kormendi and Meguire (1985) findings suggest that θ > 1. In this simple example, the mean investment ratio – the appropriate version of (23) – is γ + δk z= 1 + σy2 − σy2 . Aˆ As was pointed out in the previous section, the covariance between the investmentratio and the growth rate is positive. In this example, the appropriate version of (25) is ρ − (1 − θ )(Aˆ − δk − θ2 σy2 Aˆ 2 ) 1/2 1 + Aˆ 2 σz = σy . θ In this case, increases in σy are associated with increases (decreases) in σz if θ <(>) 1. Thus, if θ < 1, the higher the (unobserved) variance of the technology shocks (σy2 ), the higher the (measured) variances of both the growth rate, σγ2 , and the investment rate, σz2 , and the lower the mean growth rate. Moreover, in this stochastically singular setting the standard deviation of the growth rate and the investment rate are related (although not linearly). Thus, this simple model is consistent with the findings of Barlevy (2002) that the coefficient of σz is estimated to be negative, and that its introduction reduces the significance of σγ . This simple model cannot explain the apparent non-linearity in the relationship between mean and standard deviation of the growth rate process which, according to Fatás (2001), is such that the effect of σγ on γ is less negative (and can be positive) for high income countries. In order to account for this fact it is necessary to increase the degree of heterogeneity, and to consider non-linear models. Finally, the model can be reinterpreted as a multi-country model in which markets are incomplete and the distribution of the domestic shocks – the productivity shocks – is common across all countries.13 More precisely, consider a market structure in which 11 In general, this is not the case. 12 This is not necessary. In addition to differences in preferences – which we will ignore in this chapter –
ˆ δk ) as well. countries can differ in terms of (A, 13 It is possible to allow countries to share the same realization of the stochastic process. Even in this case, the demand for bonds is zero at the conjectured interest rate.
46
L.E. Jones and R.E. Manuelli
all countries can trade in a perfectly safe international bond market. In this case – which of course implies that mean growth rates are the same across countries – there is an equilibrium in which all countries choose to hold no international bonds, and the world interest rate is r ∗ = Aˆ − δk − θ σy2i Aˆ 2 . If there is a common shock that decreases the variability of every country’s technology shocks, this has a positive effect on the “world” interest rate, r ∗ , and an ambiguous impact on the world growth rate. 3.4.2. Case 2: A two sector (technology) model In the previous model, the variance of the growth rate is exogenous and equal to the variance of the technology shock. This is due, in part, to the assumption that the economy does not have another asset that can be used to diversify risk. In this section we present a very simple two-technology (or two sector) version of the model in which the variance of the growth rate is endogenously determined by the portfolio decisions of the representative agent. The main result is that, depending on the source of heterogeneity across countries, the relationship between σγ and γ need not be monotone. In particular, and depending on the source of heterogeneity across countries, the model is consistent with increases in σγ initially associated with increases in γ , and then, for large values of σγ , with decreases in the mean growth rate. To keep the model simple, we assume that the second technology is not subject to shocks, and we ignore depreciation. Thus, formally, we assume that ηb = σb = ηk = 0. However, unlike the previous case, the “safe” rate of return r satisfies A − θ σk2 < r < A. This restriction implies that α ∗ ∈ (0, 1), and guarantees that both technologies will be used to produce consumption. Since this model is a special case of the results summarized in (20) (we set the depreciation rates equal to zero for simplicity), it follows that the equilibrium mean growth rate and its variance are given by r −ρ A− r 21+ θ γ = (27a) + , θ θ σk 2 A−r 2 σγ2 = (27b) . θ σk How can we use the model to interpret the cross country evidence on variability and growth? A necessary first step is to determine which variables can potentially vary across countries. In the context of this example, a natural candidate is the vector (A, r, σk ). Before we proceed, it is useful to describe the connection between γ and σγ implied by the model. The relationship is – taking a discrete time approximation – γt =
1+θ r −ρ + σγ2 + εt , θ 2
Ch. 1: Neoclassical Models of Endogenous Growth
εt = σγ ωt ,
47
ωt ∼ N (0, 1).
It follows that if the source of cross-country differences are differences in (A, σk ) the model implies that – independently of the degree of curvature of preferences – the relationship between σγ2 and γ is positive. To see why increases in σk result in such a positive association between the two endogenous variables σγ and γ , note that, as σk rises, the economy shifts more resources to the safe technology (α ∗ decreases) and this, in turn, results in a decrease in the variance of the growth rate (which is a weighted average of the variances of the two technologies). Since the ‘risky’ technology has higher mean return than the ‘safe’ technology, the mean growth rate decreases. The reader can verify that changes in A have a similar effects. If the source of cross-country heterogeneity is due to differences in r, the implications of the model are more complex. Consider the impact of a decrease in r. From (27b) it follows that σγ2 increases and this tends to increase γ . However, as (27a) shows, this also decreases the growth rate, as it lowers the non-stochastic return. The total effect depends on the combined impact. A simple calculation shows that ∂γ 0 ⇐⇒ ∂r where
r rˆ ,
θ σk2 . 1+θ To better understand the implications of the model consider a “high” value of r; in particular, assume that r > rˆ . A decrease in r reduces σγ and, given that r > rˆ , it results in an increase in γ . Thus, for low σγ (high r) countries, the model implies a positive relationship between γ and σγ . If r < rˆ , decreases in the return to the safe technology still increase σγ , but, in this region, the growth rate decreases. Thus, in (σγ , γ ) space the model implies that, due to variations in r, the relationship between σγ and γ has an inverted U-shape. Can this model explain some of the non-linearities in the data? In the absence of further restrictions on the cross-sectional joint distribution of (A, r, σk ) the model can accommodate arbitrary patterns of association between σγ and γ . If one restricts the source of variation to changes in the return r the model implies that, for high variance countries, variability and growth move in the same direction, while for low variance countries the converse is true. If one could associate low variance countries with relatively rich countries, the implications of the model would be consistent with the type of non-linearity identified by Fatás (2001). rˆ = A −
3.4.3. Case 3: Aggregate vs. sectoral shocks The simple Ak model that we discussed in the previous section is driven by a single, aggregate, shock. In this section we consider a two sector (or two technology) economy to show that the degree of sectoral correlation of the exogenous shocks can affect the
48
L.E. Jones and R.E. Manuelli
mean growth rate. To capture the ideas in as simple as possible a model, we specialize the specification in (14) by considering the case σk = σb = σ > 0, ηb = 0,
ηk = η,
δk = δb = 0. Note that, in this setting, there is an aggregate shock, Wt , which affects both sectors (technologies) while the A sector is also subject to a specific shock, Ztk . Using the formulas derived in (18) and (20) it follows that the relevant equilibrium quantities are α∗ =
A−r , θ η2
σ2 r −ρ − (1 − θ ) + θ 2 A−r 2 σγ2 = σ 2 + . θη γ =
A−r θη
2
1+θ , 2
As before, it is useful to think of countries as indexed by (A, r, σ, η). Since changes in each of these parameters has a different impact, we analyze them separately. • An increase in σ . The increase in the standard deviation of the economy-wide shock affects both sectors equally, and it does not induce any ‘portfolio’ or sectoral reallocation of capital. The share of capital allocated to each sector (technology) is independent of σ . Since increases in σ increase σγ (in the absence of a portfolio reallocation, this is similar to the one sector case), the total effect of an increase in σ is to decrease the growth rate if 0 < θ < 1, and to increase it if θ > 1. • A decrease in r. The effect of a change in r parallels the discussion in the previous section. It is immediate to verify that a decrease in r results in an increase in σγ . However, the impact on γ is not monotonic. For high values of r, decreases in r are associated with increases in γ , while for low values the direction is reversed. Putting together these two pieces of information, it follows that the predicted relationship between σγ and γ is an inverted U-shape, with a unique value of σγ (a unique value of r) that maximizes the growth rate. • An increase in η. This change increases the ‘riskiness’ of the A technology and results in a portfolio reallocation as the representative agent decreases the share of capital in the high return sector (technology). The change implies that σγ and γ decrease. Thus, differences in η induce a positive correlation between mean and standard deviation of the growth rate. • What is the impact of differences in the degree of correlation between sectoral shocks. Note that the correlation between the two sectoral shocks is σ ν= 2 . (σ + η2 )1/2
Ch. 1: Neoclassical Models of Endogenous Growth
49
In order to isolate the impact of a change in correlation, let’s consider changes in (σ, η) such that the variance of the growth rate is unchanged. Thus, we restrict (σ, η) to satisfy A−r 2 σγ2 = σ 2 + , θη for a given (fixed) σγ . It follows that the correlation between the two shocks and the growth rate are −1 1 A−r 2 ν = 1+ , θ σ 2 (σγ2 − σ 2 ) r −ρ 1+θ 2 − σ2 + σγ . θ 2 Thus, lower correlation between sectors (in this case this corresponds to higher σ ) unambiguously lower mean growth. If countries differ in this correlation then the implied relationship between σγ and γ need not be a function; it can be a correspondence. Put it differently, the model is consistent with different values of γ associated to the same σγ . γ =
3.5. Physical and human capital In this section we study models in which individuals invest in human and physical capital. We consider a model in which the rate of utilization of human capital is constant. Even though the model is quite simple it is rich enough to be consistent with any estimated relationship between σγ and γ . We assume that output can be used to produce consumption and investment, and that market goods are used to produce human capital. This is equivalent to assuming that the production function for human capital is identical to the production function of general output. The feasibility constraints are dkt = F (kt , ht ) − δk kt − xt − ct dt + σy F (kt , ht ) dWt , dht = −δh ht + xt dt + σh ht dWt + ηht dZt , where (Wt , Zt ) is a vector of independent standard Brownian motion variables, and F is a homogeneous of degree one, concave, function. As in the previous sections, let xt = kt + ht denote total (human and non-human) wealth. With this notation, the two feasibility constraints collapse to dxt = F (αt , 1 − αt ) − δk αt + δh (1 − αt ) xt − ct dt + σy F (αt , 1 − αt )xt dWt + σh (1 − αt )xt dWt + η(1 − αt )xt dZt .
(28)
As in previous sections, the competitive equilibrium allocation coincides with the solution to the planner’s problem. The planner maximizes (13) subject to (28). The
50
L.E. Jones and R.E. Manuelli
Hamilton–Jacobi–Bellman equation corresponding to this problem is 1−θ V (x)x 2 2 c + V (x) F (α, 1 − α) − δ(α) xt − ct + σ (α) , ρV (x) = max c,α 1 − θ 2 where δ(α) = δk α + δh (1 − α), σ 2 (α) = σy2 F (α, 1 − α)2 + σh2 (1 − α)2 + η2 (1 − α)2 + σy σh F (α, 1 − α)(1 − α). 1−θ
It can be verified that a function of the form V (x) = v x1−θ solves the Hamilton– Jacobi–Bellman equation. The solution also requires that
θ ρ = θ v −1/θ + (1 − θ ) F (α, 1 − α) − δ(α) − σ 2 (α) , 2 where α is given by
θ α = arg max(1 − θ ) F (α, 1 − α) − δ(α) − σ 2 (α) . 2
For any homogeneous of degree one function F , the solution is a constant α. Moreover, α does not depend on v. Existence requires v > 0, and this is just a condition on the exogenous parameter that we assume holds.14 The growth rate and its variance are given by γ = F (α, 1 − α) − δ(α) − v −1/θ , σγ2 = σy2 F (α, 1 − α)2 + σh2 (1 − α)2 + η2 (1 − α)2 + σy σh F (α, 1 − α)(1 − α). It follows that, for the class of economies for which the planner problem has a solution (i.e. economies for which v > 0, and γ > 0), the conjectured form of V (x) solves the HJB equation, for any homogeneous of degree one function F . However, in order to make some progress describing the implications of the theory, it will prove convenient to specialize the technology and assume that F is a Cobb–Douglas function given by F (x, y) = Ax ω y 1−ω ,
0 < ω < 1.
The next step is to characterize the optimal share of wealth invested in physical capital, α, and how changes in country-specific parameters affect the mean and standard deviation of the growth rate. It turns out that the qualitative nature of the solution depends on the details of the driving stochastic process. To simplify the algebra, we assume that the human capital technology is deterministic (i.e. σh = η = 0), and that both stocks of capital (physical and human) depreciate at the same rate (δk = δh ). As 14 This is just the stochastic analog of the existence problem in endogenous growth models.
Ch. 1: Neoclassical Models of Endogenous Growth
51
indicated above, we assume that the production function is Cobb–Douglas. The first order condition for the optimal choice of α is simply φ(α)Fˆ (α) 1 − θ σy2 Fˆ (α) = 0, where Fˆ (α) ≡ Aα ω (1 − α)1−ω , ω 1−ω φ(α) = − . α 1−α The second order condition requires that −ω(1 − ω) α −2 + (1 − α)−2 Fˆ (α) 1 − θ σy2 Fˆ (α) − θ σy2 Fˆ (α)2 φ(α) < 0. Since Fˆ (α) > 0 in the relevant range, the solution is either φ(α) = 0, which corresponds to α ∗ = ω, or Fˆ (α ∗ ) = 1/θσy2 . The latter, of course, does not result in a unique α ∗ .15 The nature of the solution depends on the size of σy2 . There are two cases characterized by • Case A: σy2 ˆ1 . • Case B: σy2 >
θ F (ω) 1 . θ Fˆ (ω)
In Case A, the low variance σy2 case, the maximizer is given by α ∗ = ω, since 1 − θ σy2 Fˆ (α) > 0 for all feasible α. The second order condition is satisfied. In Case B, there are two solutions to the first order condition. They correspond to the values of α, denoted α − and α + that solve Fˆ (α ∗ ) = 1/θσy2 . By convention, let’s consider α − < ω < α + . It can be verified that in both cases the second order condition is satisfied.16 The implications of the model for the expected growth rate and its standard deviation in the two cases are Fˆ (ω) − (ρ + δ) 1 − θ 2 ˆ − σy F (ω)2 , γA = θ 2 σγA = σy Fˆ (ω), 1 1+θ 1 − (ρ + δ) , γB = θ 2 θ σy2 1 σγB = . θ σy It follows that for large σy2 , that is in Case B, the model predicts a positive relationship between mean growth and the standard deviation of the growth rate, while for small values of σy2 , Case A, the sign of the relationship depends on the magnitude of θ . 15 In the case of the Cobb–Douglas production function there are two values of α that satisfy Fˆ (α ∗ ) = 1/θσ 2 . y 16 The reader can check that, in this case, the solution α ∗ = ω does not satisfy the second order condition.
52
L.E. Jones and R.E. Manuelli
Figure 1. The mapping σγ and γ [0 < θ < 1].
Much more interesting from a theoretical point of view is the fact that the model is consistent with two countries with different σy2 to have exactly the same σγ . To see this, note that for any σγ in the range of feasible values – corresponding to the set ˆ
1/2 ] in this example – there are two values of σ , one less than ( 1 )1/2 , [0, ( F (ω) y θ ) θ Fˆ (ω) and the other greater than this threshold that result in the same σγ . The relationship between σγ and γ is a correspondence. Figure 1 displays such a relationship in the small risk aversion case, 0 < θ < 1. If the only source of cross-country heterogeneity are differences in the variability of the technology shocks, σy , the model implies that all data points should be in one of the two branches of the mapping depicted in Figure 1. By arbitrarily choosing the location of these points, the estimated relationship between σγ and γ can have any sign, and the estimated value says very little about the deep parameters of the model or, more importantly, about the effects of reducing the variability of shocks on the average growth rate. Does the nature of the result depend on the assumption 0 < θ < 1? For θ > 1 the relationship between σγ and γ is also a correspondence, and hence the model – in the absence of additional assumptions – does not pin down the sign of the correlation
Ch. 1: Neoclassical Models of Endogenous Growth
53
Figure 2. The mapping σγ and γ [θ > 1].
between σγ and γ .17 In the case of θ > 1, the size of θ matters only to determine which branch is steeper. In both cases A and B the relationship between the standard deviation of the growth rate and the mean growth rate is upward sloping. However, the low σy2 -branch is flatter (and lies above) the high σy2 -branch (see Figure 2). Since we have studied a very simple version of this class of models, it does not seem useful to determine the relevance of each branch by assigning values to the parameters. In ongoing work, we are studying more general versions of this setup. However, even this simple example suggests that some caution must be exercised when interpreting the empirical work relating the variability of the growth rate and its mean. Unless one can rule out some of these cases, theory gives ambiguous answers to the question that motivated much of the literature, i.e. do more stable countries grow faster? Moreover, the theoretical developments suggest that progress will require to estimate structural models rather than reduced form equations.
17 At this point, we have not explored what are the consequences of adding the investment output ratio to the
(theoretical) regression. However, to do this is a complete manner it seems necessary to model measurement errors, as the model is stochastically singular. We leave this for future work.
54
L.E. Jones and R.E. Manuelli
3.6. The opportunity cost view So far the models we discussed emphasize the idea that increases in the variability of the driving shocks can have positive or negative effects upon the growth rate depending on the relative importance of income and substitution effects. An alternative view is that recessions are “good times” to invest in human capital because labor – viewed as the single most important input in the production of human capital – has a low opportunity cost. In this section we present a model that captures these ideas. The model implies that the time allocated to the formation of human capital is independent of the cycle.18 It also implies that shocks to the goods production technology have no impact on growth, but that the variability of the shock process in the human capital technology decreases growth. As before, we concentrate on a representative agent with preferences described by (13). The goods production technology is given by ct + xt zt Aktα (nt ht )1−α , where nt is the fraction of the time allocated to goods production, kt is the stock of physical capital, and ht is the stock of human capital. The variable zt denotes a stationary process. To simplify the theoretical presentation we assume that capital depreciates fully. Thus, goods consumption is limited by ct zt Aktα (nt ht )1−α − kt .19 Human capital is produced using only labor in order to capture the idea that the opportunity cost of investing in human capital is market production. The technology is summarized by dht = 1 − δ + B(1 − nt ) ht dt + σh 1 − δ + B(1 − nt ) ht dWt , where, as before, Wt is a standard Brownian motion.20 Given that the problem is convex21 the competitive allocation solves the planner’s problem. It is clear that, given nt ht , physical capital will be chosen to maximize net 18 The empirical relationship between investment in human capital and the cycle is mixed. Dellas and Sakel-
laris (1997) using CPS data for all individuals aged 18 to 22 find that college enrollment id procyclical. Christian (2002) also using the CPS but restricting the sample to 18–19 years olds (so as to be able to control for family variables) finds no cyclical effects. Sakellaris and Spilimbergo (2000) study U.S. college enrollment of foreign nationals and conclude that, among those individuals coming from rich countries enrollment is countercyclical, while among students from less developed countries it is countercyclical. Moreover, college enrollment is only a partial measure of investment in human capital. Training (inside and outside business firms) is another (difficult to measure) component of increases in skill acquisition. 19 This restriction makes it possible to derive the theoretical implications of the model in a simple setting. 20 A special case of this model in which utility is assumed logarithmic, and the goods production function is not subject to shocks is analyzed in de Hek (1999). 21 Even though our choice of notation somewhat obscures this, the convexity of the technology is apparent by defining hmt = nt ht and hst = (1 − nt )ht , and adding the constraint hmt + hst ht .
Ch. 1: Neoclassical Models of Endogenous Growth
55
output. This implies that consumption is ct = A∗ zˆ t nt ht , . We guess that the relevant state where A∗ = (Aα)1/(1−α) (α −1 − 1) and zˆ t = zt variable is the vector (ˆzt , ht ), and that the value function is of the form 1/(1−α)
(ˆzt ht )1−θ . 1−θ Given this guess, the relevant Hamilton–Jacobi–Bellman equation is
A∗ 2 [ B (µ − x)ˆzh]1−θ (ˆzh)1−θ 1−θ 1−θ σh 2 ρv = max + v(ˆzh) x − v(ˆzh) θ x , x 1−θ 1−θ 2 V (ˆzt , ht ) = v
where µ ≡ 1 − δ + B, and x = 1 − δ + B(1 − n). It follows that choosing x is equivalent to choosing n. The solution to the optimization problem is given by the solution to the following quadratic equation x2 =
2(1 + µσh2 ) (1 + θ )σh2
x+
2(ρ − µ) . θ (1 + θ )σh2
In order to guarantee that utility remains bounded even in the case σh = 0 is necessary to assume that ρ − µ > 0. Simple algebra shows that the positive root of the previous equation is such that increases in σh decrease x. It follows that the stochastic process for ht is given by dht = xht dt + σh ht dWt . We now discuss the implications of the model for the growth rate of consumption (or output). Even though our results do not depend on the particular form of the zt process, it is convenient to consider the case in which zt is a geometric Brownian motion that is possibly correlated with the shock to the human capital. Specifically, we assume that dzt = zt (σw dWt + σm dMt ), where Mt is a standard Brownian motion that is uncorrelated with Wt . Ito’s lemma implies that σw2 + σm2 α α zˆ t dt + zˆ t (σw dWt + σm dMt ). 2 2 1−α (1 − α) In equilibrium, consumption (and net output) is given by dˆzt =
A∗ (µ − x)ˆzt ht . B Applying Ito’s lemma to this expression, we obtain that the growth rate of consumption dct dht dˆzt αx σh σw dt, = + + ct ht zˆ t 1−α ct =
56
or, taking a discrete time approximation, α σw2 + σm2 α σh σw + γt = x 1 + 1−α 2 (1 − α)2 α α ˜ ˜ σw + σh x Wt + σ m Mt , + 1−α 1−α γt = γ + εt , εt ∼ N 0, σγ2 , 2 2 α α σw + σh x + σm . σγ2 = 1−α 1−α
L.E. Jones and R.E. Manuelli
(29a) (29b) (29c)
Equation (29a) completely summarizes the implications of the model for the data. There are several interesting results. To simplify the notation, we will refer to Wt as the aggregate shocks and to Mt as the idiosyncratic component of the productivity shock in the goods sector. • The share of the time allocated to human capital formation – the engine of growth in this economy – is independent of the variability of the technology shock in the goods sector, as measured by (σw , σm ). • High (σw , σm ) economies are also high growth economies. Thus, if cross-country differences in σγ are mostly due to differences in (σw , σm ), the model implies a positive correlation between the standard deviation of the growth rate and mean growth. • It can be shown that increases in σh result in decreases in σh x. Thus, if countries differ in this dimension the model also implies a positive relationship between σγ and γ . • In the model, investment in physical capital (as a fraction of output) is α, independently of the distribution of the shocks. Thus, there is no sense that a regression that shows that variability does not affect the rate of investment provides evidence against the role of shocks in development. • This lack of (measured) effect on both physical and human capital investment should not be interpreted as evidence against the proposition that incentives for human or physical capital accumulation matter for growth. It is easy enough to include a tax/subsidy to the production of human capital – consider a policy that affects B – and it follows that this policy affects growth. 3.7. More on government spending, taxation, and growth In this section we consider a simple Ak model in which a government uses distortionary taxes to finance an exogenously given stochastic process for government spending. Our analysis follows Eaton (1981).22
22 For extensions of this model, see Turnovsky (1995).
Ch. 1: Neoclassical Models of Endogenous Growth
57
The representative household maximizes utility – given by (13) – by choosing consumption and saving in either capital or bonds. However, given that tax policy is exogenously fixed, it is not the case that the rate of return on bonds is risk free. On the contrary, since the government issues bonds to make up for any difference between revenue and spending it is necessary to let the return on bonds to be stochastic. The representative household problem is ∞ 1−θ −ρt ct dt F0 e max U = E (30) 1 − θ 0 subject to dkt = (rk kt − c1t ) dt + σk kt dWt ,
(31a)
dbt = (rb bt − c2t ) dt + σb bt dWt ,
(31b)
ct = c1t + c2t ,
(31c)
where kt is interpreted as capital and bt as bonds. As before, it is possible to simplify the analysis by using wealth as the state variable. Let xt ≡ kt + bt . With this notation, the single budget constraint is given by dxt = αt rk + (1 − αt )rb xt − ct dt + αt σk + (1 − αt )σb xt dWt . Since this problem is a special case of the “general” two risky assets model, it follows that the optimal solution is characterized by α=
rk −rb θ
− σb (σk − σb ) , (σk − σb )2
(32a)
b) ρ − (1 − θ )[αrk + (1 − α)rb − θ (ασk +(1−α)σ ] 2 ct = xt . θ 2
(32b)
c
The set of feasible allocations is the set of stochastic process that satisfy dkt = (Akt − ct ) dt + σ Akt dWt − dGt , dGt = gAkt dt + g σ Akt dWt . Thus, the government consumes a fraction g of the non-stochastic component of output, and a fraction g of the stochastic component. Taxes are levied on the deterministic and stochastic components of output at (possibly) different rates. The stochastic process for tax revenue is assumed to satisfy dTt = τ Akt dt + τ σ Akt dWt . In equilibrium, the parameters that determine the rate of return on capital (rk , σk ) are given by rk = (1 − τ )A,
58
L.E. Jones and R.E. Manuelli
σk = (1 − τ )σ A. The government budget constraint requires that the excess of spending over tax revenue be financed through bond issues. Thus, Bt + dGt − dTt = pt dBt , where pt Bt = bt is the value of bonds issued. The stock of capital evolves according to ct kt dt + σ (1 − g )Akt dWt . dkt = (1 − g)A − kt Note that
ct c xt 1−α = . =c =c 1+ kt kt α α
Since, in equilibrium, it must be the case that, in all states of nature, the growth rate of private wealth and the growth rate of the capital stock are the same,23 it is necessary that c αrk + (1 − α)rb − c = (1 − g)A − , (33a) α ασk + (1 − α)σb = σ (1 − g )A. (33b) The system formed by the four equations described in (32) and (33) provides the solution to the endogenous variables that need to be determined: c, α, rb , σb . It is convenient to define the excess rate of return of capital, and the excess instant variability of capital as r = rk − rb , σ = σ k − σ b . Some simple but tedious algebra shows that σγ − σk + σ , σ r = θ σ γ σ .
α=
Substituting in the remaining equations, and recalling that the instantaneous mean and standard deviation of the growth rate process is given by γ = αrk + (1 − α)rb − c, σγ = σ (1 − g )A,
23 This, of course, depends on the fact that the solution to the individual agent problem is such that bonds
and capital are held in fixed proportions.
Ch. 1: Neoclassical Models of Endogenous Growth
it follows that, (1 − τ )A − ρ − θ σγ = σ 1 − g A.
γ =
1 − θ − τ + θg σγ2 , 1 − g
59
(34a) (34b)
Equation (34a) summarizes the impact of both technology and fiscal shocks on the expected growth rate. Consider first the impact of variations in the tax regime on the relationship between the variability of the growth rate, σγ , and the average growth rate. If technology shocks, σ , are the main source of differences across countries in the standard deviation of the growth rate, then high variability countries are predicted to be low mean countries if 1 − θ − τ + θg > 0; that is, if a country has a relatively low tax rate on the stochastic component of income. This, would be the case if the base of the income tax allowed averaging over several periods. On the other hand, countries with relatively high tax rates on the random component of income display a positive relationship between the mean and the variance of their growth rates. As in more standard models, high capital income tax countries (high τ countries) have lower average growth. Differences across countries in the average size of the government, g in this notation, have no impact on growth. Finally, cross country differences in the fraction of the random component of income consumed by the government, g , induce a positive correlation between γ and σγ . This is driven by the negative impact that increases in g have on mean growth, and the equally negative effect that those changes have on σγ . Thus, high g countries display low average growth rates, which do not fluctuate much.24 3.8. Quantitative effects In this section we summarize some of the quantitative implications various models for the relationship between variability and growth. Unlike the theoretical models described above, the quantitative exercises concentrate on the role of technology shocks in models with constant – relative to output – government spending. Mendoza (1997) studies an economy in which the planner solves the following problem: max {ct }
∞ t=0
βt
ct1−θ 1−θ
subject to At+1 Rt (At − pt ct ), 24 The impact of some variables in the previous analysis differs from the results in Eaton (1981) since our
specification of the fiscal policy allows the demand for bonds (as a fraction of wealth) to be endogenous, and driven by changes in the tax code. Eaton assumes that the share of bonds, 1 − α in our notation, is given, and some tax must adjust to guarantee that demand and supply of bonds are equal.
60
L.E. Jones and R.E. Manuelli
where At is a measure of assets and pt is interpreted as the terms of trade. Since this equation can be rewritten as kt+1 = rt+1 (kt − ct ), where kt = At /pt is the stock of assets (capital) measured in units of consumption, and rt+1 = Rt pt /pt+1 is the random rate of return, it is clear that Mendoza’s model is a stochastic version of an Ak model. The rate of return, rt is assumed to be lognormally distributed with mean and variance given by µr = eµ+σ σr2 = µ2r e
2 /2
,
σ 2 −1
.
It follows that the standard deviation of the growth rate is σγ = σ . Mendoza studies the effect of changing σ from 0 to 0.15, holding µr constant. To put the exercise in perspective, the average across countries of the standard deviations of the growth rate or per capita output in the Summers–Heston dataset is 0.06. Thus, the model is calibrated at a fairly high level of variability. The results depend substantially on the assumed value of θ . For θ = 1/2, the non-stochastic growth rate is 3.3%. If σ = 0.10, it decreases to 2.5%, while it is 1.6% when σ = 0.15. For θ = 2.33 (Mendoza’s preferred specification), the growth rate increases from 0.7% to 0.9% in the given range. For other values of the coefficient of risk aversion, the impact of uncertainty is also small. In summary, unless preferences are such that the degree of intertemporal substitution is large, increases in rate of return uncertainty have a small impact on mean growth rates. Jones et al. (2003) analyze the following planner’s problem:
t 1−θ max Et β ct v(t )/(1 − θ ) , t
subject to ct + xzt + xht + xkt F (kt , zt , st ), zt M(nzt , ht , xzt ), kt+1 (1 − δk )kt + xkt , ht+1 (1 − δh )ht + G(nht , ht , xht ), t + nht + nzt 1. For their quantitative exercise, they specify the following functional forms: nh = xz = 0, v() =
ψ(1−σ )
nz = n, ,
F (k, z, s) = sAk α z1−α , G(h, xh ) = xh ,
Ch. 1: Neoclassical Models of Endogenous Growth
61
M(n, h) = nh, st = exp ζt − ζt+1
σε2 , 2(1 − ρ 2 ) = ρζt + εt+1 .
The model is calibrated to match the average growth, in a cross section of countries, and its standard deviation. Jones et al. (2003) consider the impact of changing the standard deviation of the shock, st , from 0 to 0.15. The impact on the growth rate depends on the curvature of the utility function. For preferences slightly less curved than the log, the model predicts an increase in the growth rate of 0.7% an annual basis, while for θ = 1.5, the effects is an increase in the growth rate of 0.25%. However, the model predicts that σγ – which is endogenous – is unusually high (of the order of 0.10) unless θ 1.5. Thus, Jones et al. (2003) obtain results that are quite different from those of Mendoza (1997). There are two important differences between the models: First, while Mendoza (1997) assumes that shocks are i.i.d., Jones et al. (2003) set the first order correlation parameter, ρ, to 0.95. Second, while Mendoza assumes a constant labor supply, Jones et al. allow for the number of hours to vary with the shock. In order to disentangle the effect of the components of the standard deviation of the technology shock, Jones et al. vary σε and ρ in a series of experiments, where σs = σε /(1 − ρ 2 )1/2 , where σε is the standard deviation of the innovations. They find that changes in σε appear quantitatively more important than those and ρ. Moreover, they also find that the relative variability of hours worked is very sensitive to the precise value of θ . Economies with high θ , and lower σγ in their specification, also display substantially less variability in hours worked. Even though it is not possible to determine on the basis of these two results which is the critical feature that accounts for the differences between the results obtained by Mendoza (1997) and Jones et al. (2003), it seems that the assumption of a flexible supply of hours – which determines the rate of utilization of human capital – is a leading candidate.25 In a series of papers, Krebs (2003a, 2003b) explores the impact of changes in uncertainty in models where markets are incomplete. Building on the work of de Hek (1999), Krebs (2003a) studies the impact of shocks to the depreciation rate of the capital stock. Even though he assumes that instantaneous utility is logarithmic, he finds that increases in the standard deviation of the shock, decrease growth rates. This result is driven by the “location” of the shocks, and it does not require market incompleteness.26 Quantitatively, Krebs (2003a) finds that an increase in the standard deviation of the growth rate from 0 to 0.15 (a fairly large value relative to the world average) reduces growth 25 In subsequent work, Jones et al. (2003) analyze the business cycle properties of the same class of models,
and they show that are capable of generating higher serial correlation in the growth rate of output than similar exogeneous growth models. 26 de Hek (1999) shows theoretically that increases in the variance of the depreciation shock decrease average growth even if markets are complete, and the shocks are aggregate shocks.
62
L.E. Jones and R.E. Manuelli
from 2.13% to 2.00%. If the variability is increased to 0.20 the growth rate drops to 1.5%. However, these values are substantially higher than those observed in international data.27 The theoretical analysis of the impact of different forms of uncertainty on the growth of the economy is still in its infancy. Simple existing models suggest that the sign of the relationship between the variability of a country’s growth rate and its average growth rate is ambiguous. Thus, theoretical models that restrict more moments can help in understanding the effect of fluctuations on growth and welfare. The few quantitative studies that we reviewed have produced conflicting results. It seems that the precise nature of the shocks, their serial correlation properties and the elasticity of hours with respect to shocks all play a prominent role in accounting for the variance in predicted outcomes. Much more work is needed to identify realistic and tractable models that will be capable of confronting both time series and cross country observations.
4. Concluding comments In this chapter we briefly presented the basic insights about the growth process that can be learned from studying standard convex models with perfectly functioning markets. We emphasized three aspects of those models. First, the impact of fiscal policy on growth. A summary of the current state of knowledge is that theoretical models have ambiguous implications about the effect of taxes on growth. The key feature is the importance of market goods in the production of human capital. If, as Lucas (1988) assumes, no market goods are needed to produce new human capital, the impact of income taxes on growth is small (or zero in some cases). If, on the other hand, market goods are necessary to produce human capital then taxes play a more important role, and they have a large impact on growth. It seems that the next step is to use detailed models of the process of human capital formation and to explore the implications that they have for the age-earnings profile to identify the parameters of the production function of human capital. A first step in that direction is in Manuelli and Seshadri (2005). A second important issue that features prominently in the discussion of the relative merits of convex and non-convex models is the role of innovation. The standard argument claims that innovation is a one-off investment (with low copying costs) and hence that this technology is inconsistent with price taking behavior. In this chapter we elaborate on the ideas discussed by Boldrin and Levine (2002) and show that it is possible to reconcile the existence of a non-convexity with competitive behavior. The last major theme covered in this survey is the relationship between fluctuations and growth. An important question is whether technological or policy induced fluctuations affect the growth rate of an economy. This is relevant for the time series experience
27 Krebs (2003a) does not have an aggregate shock. His model predicts that aggregate growth is constant. He
calibrates his model to match the standard deviation of the growth rate of individual income.
Ch. 1: Neoclassical Models of Endogenous Growth
63
of a single country (e.g., the discussion about the role of post-war stabilization policies on the growth rate of the American economy), as well as the prescriptions of international agencies for national policies. We discuss the empirical evidence and find it conflicting. It is not easy to identify a clear pattern between fluctuations and growth. To shed light on why this might be the case, we discuss a series of theoretical models. We show that the relationship between the growth rate and its standard deviation has an ambiguous sign. We also describe more precisely how one might identify the parameters of preferences and technologies that determine the sign of the relationship. This is one area of research in which more theoretical and empirical work will have a high marginal value.
Acknowledgements Both authors acknowledge the generous support of the National Science Foundation.
References Aghion, P., Howitt, P. (1998). Endogenous Growth Theory. MIT Press, Cambridge, MA. Aizenman, J., Marion, N. (1999). “Volatility and investment”. Economica 66 (262). Alvarez, F., Stokey, N.L. (1998). “Dynamic programming with homogeneous functions”. Journal of Economic Theory 82 (1), 167–189. Barlevy, G. (2002). “The cost of business cycle under endogenous growth”. Working Paper. Northwestern University. Barro, R. (1990). “Government spending in a simple model of economic growth”. Journal of Political Economy, Part 2 98 (5), S103–S125. Barro, R.M., Sahasakul, C. (1986). “Measuring average marginal tax rates from social security and the individual income tax”. Journal of Business 59 (4), 555–566. Barro, R., Sala-i-Martin, X. (1995). Economic Growth. McGraw-Hill, New York, Saint Louis. Boldrin, M., Levine, D. (2002). “Perfectly competitive innovation”. Staff Report 303. Federal Reserve Bank of Minneapolis. Bond, E., Ping, W., Yip, C.K. (1996). “A general two sector model of endogenous growth with human and physical capital: Balanced growth and transitional dynamics”. Journal of Economic Theory 68 (1), 149– 173. Brock, W., Mirman, L. (1972). “Optimal economic growth under uncertainty: The discounted case”. Journal of Economic Theory 4, 497–513. Cass, D. (1965). “Optimum growth in an aggregative model of capital accumulation”. Review of Economic Studies 32, 233–240. Christian, M.S. (2002). “Liquidity constraints and the cyclicality of college enrollment”. Working Paper. University of Michigan. Dawson, J.W., Stephenson, E.F. (1997). “The link between volatility and growth: Evidence from the states”. Economics Letters 55, 365–369. Debreu, G. (1954). “Valuation equilibria and Pareto optimum”. Proceedings of the National Academy of Sciences 40 (7), 588–592. de Hek, P.A. (1999). “On endogenous growth under uncertainty”. International Economic Review 40 (3), 727–744.
64
L.E. Jones and R.E. Manuelli
de Hek, P., Roy, S. (2001). “Sustained growth under uncertainty”. International Economic Review 42 (3), 801–813. Dellas, H., Sakellaris, P. (1997). “On the cyclicality of schooling: Theory and evidence”. Working Paper. University of Maryland. Eaton, J. (1981). “Fiscal policy, inflation and the accumulation of risky capital”. Review of Economic Studies XLVIII, 435–445. Fatás, A. (2001). “The effect of business cycle on growth”. Working Paper. INSEAD. Grier, K.B., Tullock, G. (1989). “An empirical analysis of cross-national economic growth, 1951–1980”. Journal of Monetary Economics 24 (2), 259–276. Jones, L.E., Manuelli, R.E. (1990). “A convex model of equilibrium growth: Theory and policy implications”. Journal of Political Economy 98, 1008–1038. Jones, L.E., Manuelli, R.E. (1997). “The sources of growth”. Journal of Economic Dynamics and Control 27, 75–114. Jones, L.E., Manuelli, R.E., Rossi, P.E. (1993). “Optimal taxation in models of endogenous growth”. Journal of Political Economy 101 (3), 485–517. Jones, L.E., Manuelli, R.E., Siu, H. (2003). “Fluctuations in convex models of endogenous growth II: Business cycle properties”. Working Paper. Jones, L.E., Manuelli, R.E., Siu, H., Stacchetti, E. (2003). “Fluctuations in convex models of endogenous growth I: Growth effects”. Working Paper. Kaganovich, M. (1998). “Sustained endogenous growth with decreasing returns and heterogeneous capital”. Journal of Economic Dynamics and Control 22 (10), 1575–1603. Koopmans, T. (1965). “On the concept of optimal economic growth”. In: The Economic Approach to Development Planning. North-Holland, Amsterdam. Kormendi, R.L., Meguire, P.G. (1985). “Macroeconomic determinants of growth: Cross-country evidence”. Journal of Monetary Economics 16 (September), 141–163. Krebs, T. (2003a). “Human capital risk and economic growth”. Quarterly Journal of Economics 118, 709–745. Krebs, T. (2003b). “Growth and welfare effects of business cycles in economies with idiosyncratic human capital risk”. Review of Economic Dynamics 6 (4), 846–868. Kroft, K., Lloyd-Ellis, H. (2002). “Further cross-country evidence on the link between growth, volatility and business cycles”. Working Paper. Ladron de Guevara, A., Ortigueira, S., Santos, M.S. (1997). “Equilibrium dynamics in two-sector models of endogenous growth”. Journal of Economic Dynamics and Control 21 (1), 115–143. Leland, H. (1974). “Optimal growth in a stochastic environment”. Review of Economic Studies 41 (1), 75–86. Levhari, D., Srinivasan, T.N. (1969). “Optimal savings under uncertainty”. Review of Economic Studies 36 (106), 153–163. Li, W., Sarte, P. (2001). “Growth effects of progressive taxes”. Working Paper. Federal Reserve Bank of Richmond. Lucas, R.E. Jr. (1988). “On the mechanics of economic development”. Journal of Monetary Economics 22, 3–42. Lucas, R.E. Jr. (1990). “Supply-side economics: An analytical review”. Oxford Economics Papers 42, 293– 316. Manuelli, R., Seshadri, A. (2005). “Human capital and the wealth of nations”. Working Paper. University of Wisconsin. Martin, P., Rogers, C.A. (2000). “Long term growth and short term economic instability”. European Economic Review 2 (44), 359–381. McGrattan, E., Prescott, E.C. (2003). “Average debt and equity returns: Puzzling?” Staff Report 313. Federal Reserve Bank of Minneapolis. McGrattan, E., Prescott, E.C. (2004) (revised). “Taxes, regulations, and the value of U.S. and U.K. Corporations”. Staff Report 309. Federal Reserve Bank of Minneapolis. Mendoza, E. (1997). “Terms of trade uncertainty and economic growth”. Journal of Development Economics 54, 323–356.
Ch. 1: Neoclassical Models of Endogenous Growth
65
Mulligan, C. (2003). “Capital tax incidence: Fisherian impressions from the time series”. NBER Working Paper 9916. Phelps, E.S. (1962). “The accumulation of risky capital: A sequential utility analysis”. Econometrica 30, 729–743. Ramey, G., Ramey, V. (1995). “Cross-country evidence on the link between volatility and growth”. American Economic Review 85, 1138–1151. Rebelo, S. (1991). “Long run policy analysis and long run growth”. Journal of Political Economy 99, 500– 521. Sakellaris, P., Spilimbergo, A. (2000). “Business cycle and investment in human capital: International evidence on higher education”. Carnegie-Rochester Series on Economic Policy 52, 221–256. Shell, K. (1967). “A model of inventive activity and capital accumulation”. In: Shell, K. (Ed.), Essays on the Theory of Optimal Economic Growth. MIT Press, Cambridge, MA. Shell, K. (1973). “Inventive activity, industrial organization and economic activity”. In: Mirrlees, J., Stern, N. (Eds.), Models of Economic Growth. Macmillan, London, UK. Siegler, M.V. (2001). “International growth and volatility in historical perspective”. Working Paper. Department of Economics, Williams College. November. Solow, R.M. (1956). “A contribution to the theory of economic growth”. Quarterly Journal of Economics 70, 65–94. Stokey, N., Rebelo, S. (1995). “Growth effects of flat-tax rates”. Journal of Political Economy 103, 519–550. Turnovsky, S.J. (1995). Methods of Macroeconomic Dynamics. MIT Press, Cambridge, MA; London, UK. Uzawa, H. (1964). “Optimal growth in a two sector model of capital accumulation”. Review of Economic Studies 31, 1–24.
Chapter 2
GROWTH WITH QUALITY-IMPROVING INNOVATIONS: AN INTEGRATED FRAMEWORK PHILIPPE AGHION Harvard University PETER HOWITT Brown University
Contents Abstract Keywords 1. Introduction 2. Basic framework 2.1. A toy version of the Aghion–Howitt model 2.2. A generalization 2.3. Alternative formulations 2.4. Comparative statics on growth
3. Linking growth to development: convergence clubs 3.1. A model of technology transfer 3.2. World growth and distribution 3.3. The role of financial development in convergence 3.4. Concluding remark
4. Linking growth to IO: innovate to escape competition 4.1. The theory 4.2. Empirical predictions 4.3. Empirical evidence and relationship to literature 4.4. A remark on inequality and growth
5. Scale effects 5.1. Theory 5.1.1. The Schumpeterian (fully endogenous) solution 5.1.2. The semi-endogenous solution 5.2. Evidence 5.2.1. Results 5.3. Concluding remarks
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01002-6
68 68 69 69 69 71 75 75 76 78 81 81 84 84 86 89 89 92 92 92 93 93 94 95 97
68
P. Aghion and P. Howitt
6. Linking growth to institutional change 6.1. From Schumpeter to Gerschenkron 6.2. A simple model of appropriate institutions 6.3. Appropriate education systems 6.3.1. Distance to frontier and the growth impact of higher education 6.3.2. Empirical evidence
7. Conclusion References
98 98 100 101 102 105 106 107
Abstract In this chapter we argue that the endogenous growth model with quality-improving innovations provides a framework for analyzing the determinants of long-run growth and convergence that is versatile, simple and empirically useful. Versatile, as the same framework can be used to analyze how growth interacts with development and crosscountry convergence and divergence, how it interacts with industrial organization and in particular market structure, and how it interacts with organizations and institutional change. Simple, since all these aspects can be analyzed using the same elementary model. Empirically useful, as the framework generates a whole range of new microeconomic and macroeconomic predictions while it addresses empirical criticisms raised by other endogenous growth models in the literature.
Keywords growth, innovation, convergence, competition, scale effect, institutions JEL classification: O1, O31, O40
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
69
1. Introduction Technological progress, the mainspring of long-run economic growth, comes from innovations that generate new products, processes and markets. Innovations in turn are the result of deliberate research and development activities that arise in the course of market competition. These Schumpeterian observations constitute the starting point of that branch of endogenous growth theory built on the metaphor of quality improvements, whose origins lie in the partial-equilibrium industrial-organization literature on patent races. Our own entry to that literature was the pre-publication version of Chapter 10 of Tirole (1988). We argued in Aghion and Howitt (1998) that by using Schumpeter’s insights to develop a growth model with quality-improving innovations one can provide an integrated framework for understanding not only the macroeconomic structure of growth but also the many microeconomic issues regarding incentives, policies, and institutions that interact with growth. Who gains from innovations, who loses, and how, much all depend on institutions and policies. By focusing on these influences in a model where entrepreneurs introduce new technologies that render previous technologies obsolete we hope to understand why those who would gain from growth prevail in some societies, while in others they are blocked by those who would lose. In this chapter we show that the growth model with quality-improving innovations (also referred to as the “Schumpeterian” growth paradigm) is not only versatile but also simple and empirically useful. We illustrate its versatility by showing how it sheds light on such diverse issues as cross-country convergence, the effects of product-market competition on growth, and the interplay between growth and the process of institutional change. We illustrate its simplicity by building our analysis around an elementary discrete-time model. We illustrate its empirical usefulness by summarizing recent papers and studies that test the microeconomic and macroeconomic implications of the framework and that address what might seem like empirically questionable aspects of the earliest prototype models in the literature. The paper is organized as follows. Section 2 develops the basic framework. Section 3 uses it to analyze convergence and divergence patterns in the cross-country data. Section 4 analyses the interaction between growth and product market competition. Section 5 deals with the scale effect of growing population. Section 6 analyzes the interplay between institutional change and technological change, and Section 7 provides some concluding remarks and suggestions for future research.
2. Basic framework 2.1. A toy version of the Aghion–Howitt model Asked by colleagues to show them the simplest possible version of the quality-ladder model of endogenous growth which they could teach to second year undergraduate stu-
70
P. Aghion and P. Howitt
dents, we came out with the following stripped-down version of Aghion and Howitt (1992). Time is discrete, indexed by t = 1, 2, . . . , and at each point in time there is a mass L of individuals, each endowed with one unit of skilled labor that she supplies inelastically. Each individual lives for one period and thus seeks to maximize her consumption at the end of her period. Each period a final good is produced according to the Cobb–Douglas technology: y = Ax α ,
(1)
where x denotes the quantity of intermediate input used in final good production, and A is a productivity parameter that reflects the current quality of the intermediate good. The intermediate good is itself produced using labor according to a simple one-forone technology, with one unit of labor producing one unit of the current intermediate good. Thus x also denotes the amount of labor currently employed in manufacturing. But labor can also be employed in research to generate innovations. Each innovation improves the quality of the intermediate input, from A to γ A where γ > 1 measures the size of the innovation. Innovations result from research investment. More specifically, there is an innovator who, if she invests z units of labor in research, innovates with probability λz and thereby discover an improved version of the intermediate input. The innovator enjoy monopoly power in the production of the intermediate good, but faces a competitive fringe who can produce a unit of the same intermediate good by using χ > 1 units of labor instead of one. For χ < 1/α, this competitive fringe is binding which means that χwt is the maximum price the innovator can charge without being driven out of the market. Her profit is thus equal to: πt = (χ − 1)wt xt , where wt xt is the wage bill. This monopoly rent, however, is assumed to last for one period only, after which imitation allows other individuals to produce intermediate goods of the same quality. The model is entirely described by two equations. The first is a labor market clearing equation, which states that at each period total labor supply L is equal to manufacturing labor demand x plus total research labor n, that is: L = xt + nt for all t. The second is a research arbitrage equation which says that in equilibrium at any date t the amount of research undertaken by the innovator must equate the marginal cost of a unit of research labor with the expected marginal benefit. The marginal cost is just the manufacturing wage wt . The expected benefit comes from raising the probability of success by λ · 1 = λ, in which case she earns the monopoly profit πt involved in producing the intermediate good for the final good sector. Thus the research arbitrage equation can be expressed as: wt = λγ πt
(research arbitrage)
where the factor γ on the right-hand side of the equation, simply stems from the fact that an innovation multiplies wages and profits by γ .
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
71
Using the fact that the allocation of labor between research and manufacturing remains constant in steady-state, we can drop time subscripts. Then, substituting for πt in the research arbitrage equation, dividing through by w, and using the labor market clearing equation to substitute for x, we obtain: 1 = λγ (χ − 1)(L − n) which solves for the steady-state amount of research labor, namely: n=L−
1 . λγ (χ − 1)
The equilibrium expected rate of productivity growth in steady-state, is then simply given by: g = λn(γ − 1) and it therefore depends upon the characteristics of the economic environment as described by the parameters λ, γ , χ, and L. In Section 2.3 we interpret the comparative statics of growth with respect to all these parameters, and suggest preliminary policy conclusions. The model is extremely simple, although at the cost of making some oversimplifying assumptions. In particular, we assumed only one intermediate sector, and that labor is the only input into research. In the next sections we relax these two assumptions. We develop a slightly more elaborated version of the quality-ladder model that we then extend in several directions to capture important aspects of the growth and development process. 2.2. A generalization There are three kinds of goods in the economy: a general-purpose good, a large number m of different specialized intermediate inputs, and labor. Time is discrete, indexed by t = 1, 2, . . . , and there is a mass L of individuals, each endowed with one unit of skilled labor that she supplies inelastically.1 The general good is produced competitively using intermediate inputs and labor, according to the production function: yt = Σ1m Ait1−α xitα (L/m)1−α (2)
1 The model we present here is a simplified discrete-time version of the Aghion and Howitt (1992) model of creative destruction, which draws upon Acemoglu, Aghion and Zilibotti (2002). Grossman and Helpman (1991) presented a variant of the framework in which the x’s are final consumption goods and utility is loglinear. An early attempt at developing a Schumpeterian growth model with patent races in deterministic terms was presented by Segerstrom et al. (1990). Corriveau (1991) developed an elegant discrete-time model of growth through cost-reducing innovations.
72
P. Aghion and P. Howitt
where each xit is the flow of intermediate input i used at date t, and Ait is a productivity variable that measures the quality of the input. The general good is used in turn for consumption, research, and producing the intermediate inputs. The expected growth rate of any given productivity variable Ait is: g = E(Ait /Ai,t−1 ) − 1.
(3)
There is no i subscript on g because, as we shall see, all sectors are ex ante identical and hence will have the same productivity-growth rate. Likewise there is no t subscript because, as we shall also see, the system will go immediately to a constant steady-state expected growth rate. Productivity growth in any sector i results from innovations, which create improved versions of that intermediate input. More precisely, each innovation at t multiplies the pre-existing productivity parameter Ai,t−1 of the best available input by a factor γ > 1. Innovations in turn result from research. If Nit units of the general good are invested at the beginning of the period, some individual can become the new “leading-edge” producer of the intermediate input with probability µit , where:2 µit = λf (nit ),
f > 0, f < 0, f (0) = 0,
it and nit ≡ γ ANi,t−1 is productivity-adjusted R&D expenditure in the sector. We divide by γ Ai,t−1 , the targeted productivity parameter, to take into account the “fishing-out” effect – on average each quality improvement is harder to bring about than the previous one. Assume the time period is short enough that we may ignore the possibility of more than one successful innovator in the same sector. Then: γ Ai,t−1 with probability λf (nit ), Ait = (4) with probability 1 − λf (nit ). Ai,t−1
According to (3) and (4) the expected productivity-growth rate in each sector can be expressed as the product of the frequency of innovations λf (n) and the incremental size of innovations (γ − 1): g = λf (n)(γ − 1)
(5)
in an equilibrium where productivity-adjusted research is the same constant n in each sector. We assume moreover that the outcome of research in any one sector is statistically independent of the outcome in every other sector. The model determines research n, and therefore the expected productivity-growth rate g, using a research arbitrage equation that equates the expected cost and benefit 2 More precisely, f (n) = F (n, k) where k is some specialized research factor in fixed supply and F is a constant-returns function. Since there is free entry in research, the equilibrium price of k adjusts so that the expected profit of an R&D firm is zero. Since this price plays no role in the analysis of growth we suppress the explicit representation of k and deal only with the decreasing-returns function f . (Of course the constantreturns assumption can be valid only over some limited range of inputs, since F is bounded above by unity.)
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
73
of research. The payoff to research in any sector i is the prospect of a monopoly rent πit if the research succeeds in producing an innovation. This rent lasts for one period only, as all individuals can imitate the current technology next period. Hence the expected benefit from spending one unit on research is πit times the marginal probability λf (n)/(γ Ai,t−1 ): 1 = λf (n) πit /(γ Ai,t−1 ) . To solve this equation for n we need to determine the productivity-adjusted monopoly rent πit /Ait to a successful innovator. As before, we assume that this innovator can produce the leading-edge input at a constant marginal cost of one unit of the general good. But she faces a competitive fringe of imitators who can produce the same product at higher marginal cost χ, where χ ∈ (1, 1/α)3 is an inverse measure of the degree of product market competition or imitation in the economy.4 Thus her monopoly rent is again equal to: πit = (pit − 1)xit = (χ − 1)xit .
(6)
A monopolist’s output xit will be the amount demanded by firms in the general sector when faced with the price χ; that is, the quantity such that χ equals the marginal product of the ith intermediate good in producing the general good: χ = ∂yt /∂xit = α(mxit /Ait L)α−1 .
(7)
Hence: πit = δ(χ)Ait L/m
(8)
where 1
δ(χ) ≡ (χ − 1)(χ/α) α−1 ,
δ (χ) > 0.5
Therefore we can write the research arbitrage equation, taking into account that γ Ai,t−1 = Ait because a monopolist is someone who has just innovated, as: 1 = λf (n)δ(χ)L/m
(9)
which we assume in this section has a positive solution. 3 It is easily verified that if there were no fringe then the unconstrained monopolist would charge a price
equal to 1/α, but at that price the fringe could profitably undercut her because its unit cost is χ < 1/α. 4 If no innovation occurs then some firm will produce, but with no cost advantage over the fringe because everyone is able to produce last period’s intermediate input at a constant marginal cost of unity. 5 To see that δ > 0 note that: χ
χ 1 d ln(δ(x)) = − >0 dχ χ −1 1−α
where the last inequality follows from the assumption that χ < 1/α.
74
P. Aghion and P. Howitt
The expected productivity growth rate is determined by substituting the solution of (9) into the growth equation (5). In the special case where the research-productivity function f takes the simple form: √ f (n) = 2n, we have: g = λ2 δ(χ)(L/m)(γ − 1).
(10)
As it turns out, g is not only the expected growth rate of each sector’s productivity parameter but also the approximate growth rate of the economy’s per-capita GDP. This is because per-capita GDP is approximately proportional to the unweighted average of the sector-specific productivity parameters:6 1 Ait . m m
At =
i=1
6 To see this, note that GDP equals the sum of value added in the general sector and in the monopolized intermediate sectors. (There is no value added in the competitive intermediate sectors because their output is priced at the cost of the intermediate inputs. Also, we follow the standard national income accounting practice of ignoring the output (patents) of the research sector.) According to (7) the output of each monopolized sector (i ∈ Mt ) at t is: 1
xit = (α/χ) 1−α (L/m)Ait
(i ∈ Mt ).
The output of each competitive sector (i ∈ Ct ) at t is the amount demanded when its price is unity; setting χ = 1 in (7) yields: 1
xit = (α) 1−α (L/m)Ait
(i ∈ Ct ).
Substituting these into (2) and rearranging yields the following expression for per-capita general output: α α 1 1 #Mt #Ct yt /L = (α/χ) 1−α Ait + α 1−α Ait m #Mt m #Ct i∈Mt
i∈Ct
where #Mt is the number of monopolized sectors and #Ct the number of competitive sectors. By the law of large numbers, the fraction of sectors #Mt /m monopolized, i.e. the fraction in which there was an innovation at t, is approximately the probability of success in research in each sector: µ = λf (n), and the fraction is approximately 1 − µ. The average productivity parameter among of sectors #Ct /m that are competitive 1 A ) is just γ times the average productivity parameter of those sectors monopolized sectors ( #M it i∈M t t last period; since innovations are spread randomly across sectors this is approximately γ times the average across all sectors last period: γ At−1 . Likewise the average productivity parameter across competitive sectors is the average across sectors that did not innovate, which is approximately At−1 . Making these substitutions into the above expression for per-capita general output yields: α α yt /L (α/χ) 1−α µγ + α 1−α (1 − µ) At−1 ≡ ζ At−1 .
Since labor is paid its marginal product in the general sector, the wage rate is: wt = ∂yt /∂L = (1 − α)yt /L (1 − α)ζ At−1
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
75
Since (a) all sectors have an expected growth rate of g, (b) the sectoral growth rates are statistically independent of each other and (c) there is a large number of them, therefore the law of large numbers implies that the average grows at approximately the same rate g as each component. 2.3. Alternative formulations There are many other ways of formulating the basic model. We note two of them here for future reference. In the first one, as in the above toy model, the general good is used only for consumption, while skilled labor is the only factor used in producing intermediate products and research. The general good is produced by the intermediate inputs in combination with a specialized factor (for example unskilled labor) available in fixed supply. In this formulation, the growth equation is the same as (5) above, but with n being interpreted as the amount of skilled labor allocated to R&D. This version will be spelled out in somewhat more detail in Section 5. The other popular version is one with intersectoral spillovers, in which each innovation produces a new intermediate product in that sector embodying the maximum At−1 of all productivity parameters of the last period, across all sectors, times some factor γ that depends on the flow of innovations in the whole economy. The idea here is that if a sector has been unlucky for a long time, while the rest of the economy has progressed, the technological progress elsewhere spills over into the innovation in this sector, resulting in a larger innovation than if the innovation had occurred many years ago. The model in Section 3 is a variant of this version. 2.4. Comparative statics on growth Equation (10) delivers several comparative-statics results, each with important policy implications on how to “manage” the growth process: 1. Growth increases with the productivity of innovations λ and with the supply of skilled labor L: both results point to the importance of education, and particularly higher education, as a growth-enhancing factor. Countries that invest more in higher education will achieve a higher productivity of research activities and also reduce the opportunity cost of R&D by increasing the aggregate supply of skilled labor. An increase in the size of population should also bring about an increase in growth by raising L. This “scale effect” has been challenged in the literature and will be discussed in Section 5. which is also per-capita value-added in the general sector. By similar reasoning, (8) implies that per-capita value added in monopolized intermediate sectors is: 1 #Mt δ(χ)Ait L/m = δ(χ) Ait δ(χ)µγ At−1 . (1/L) m #Mt i∈Mt
i∈Mt
Therefore each component of per-capita GDP is approximately proportional to At−1 . Since At grows at approximately the constant rate g therefore per-capita GDP is approximately proportional to At .
76
P. Aghion and P. Howitt
2. Growth increases with the size of innovations, as measured by γ . This result points to the existence of a wedge between private and social innovation incentives. That is, a decrease in size would reduce the cost of innovation in proportion to the expected rents; the research arbitrage equation (9) shows that these two effects cancel each other, leaving the equilibrium level of R&D independent of size. However, Equation (10) shows that the social benefit from R&D, in the form of enhanced growth, is proportional not to γ but to the “incremental size” γ − 1. When γ is close to one it is not socially optimal to spend as much on R&D as when γ is very large, because there is little social benefit; yet a laissez-faire equilibrium would result in the same level of R&D in both cases. 3. Growth is decreasing with the degree of product market competition and/or with the degree of imitation as measured inversely by χ. Thus patent protection (or, more generally, better protection of intellectual property rights), will enhance growth by increasing χ and therefore increasing the potential rewards from innovation. However, pro-competition policies will tend to discourage innovation and growth by reducing χ and thereby forcing incumbent innovators to charge a lower limit price. Existing historical evidence supports the view that property rights protection is important for sustained long-run growth; however the prediction that competition should be unambiguously bad for innovations and growth is questioned by all recent empirical studies, starting with the work of Nickell (1996) and Blundell, Griffith and Van Reenen (1999). In Section 4 we shall argue that the Schumpeterian framework outlined in this section can be extended so as to reconcile theory and evidence on the effects of entry and competition on innovations, and that it also generates novel predictions regarding these effects which are borne out by empirical tests.
3. Linking growth to development: convergence clubs With its emphasis on institutions, the Schumpeterian growth paradigm is not restricted to dealing with advanced countries that perform leading-edge R&D. It can also shed light on why some countries that were initially poor have managed to grow faster than industrialized countries, whereas others have continued to fall further behind. The history of cross-country income differences exhibits mixed patterns of convergence and divergence. The most striking pattern over the long run is the “great divergence” – the dramatic widening of the distribution that has taken place since the early 19th Century. Pritchett (1997) estimates that the proportional gap in living standards between the richest and poorest countries grew more than five-fold from 1870 to 1990, and according to the tables in Maddison (2001) the proportional gap between the richest group of countries and the poorest7 grew from 3 in 1820 to 19 in 1998. But over 7 The richest group was Western Europe in 1820 and the “Euopean Offshoots” (Australia, Canada, New Zealand and the United States) in 1998. The poorest group was Africa in both years.
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
77
the second half of the twentieth century this widening seems to have stopped, at least among a large group of nations. In particular, the results of Barro and Sala-i-Martin (1992), Mankiw, Romer and Weil (1992) and Evans (1996) seem to imply that most countries are converging to parallel growth paths. However, the recent pattern of convergence is not universal. In particular, the gap between the leading countries as a whole and the very poorest countries as a whole has continued to widen. The proportional gap in per-capita income between MayerFoulkes’s (2002) richest and poorest convergence groups grew by a factor of 2.6 between 1960 and 1995, and the proportional gap between Maddison’s richest and poorest groups grew by a factor of 1.75 between 1950 and 1998. Thus as various authors8 have observed, the history of income differences since the mid 20th century has been one of “club-convergence”; that is, all rich and most middle-income countries seem to belong to one group, or “convergence club”, with the same long-run growth rate, whereas all other countries seem to have diverse long-run growth rates, all strictly less than that of the convergence club. The explanation we develop in this section for club convergence follows Howitt (2000), who took the cross-sectoral-spillovers variant of the closed-economy model described in Section 2.3 and allowed the spillovers to cross international as well as intersectoral borders. This international spillover, or “technology transfer”, allows a backward sector in one country to catch up with the current technological frontier whenever it innovates. Because of technology transfer, the further behind the frontier a country is initially, the bigger the average size of its innovations, and therefore the higher its growth rate for a given frequency of innovations. As long as the country continues to innovate at some positive rate, no matter how small, it will eventually grow at the same rate as the leading countries. (Otherwise the gap would continue to rise and therefore the country’s growth rate would continue to rise.) However, countries with poor macroeconomic conditions, legal environment, education system or credit markets will not innovate in equilibrium and therefore they will not benefit from technology transfer, but will instead stagnate. This model reconciles Schumpeterian theory with the evidence to the effect that all but the poorest countries have parallel long-run growth paths. It implies that the growth rate of any one country depends not on local conditions but on global conditions that impinge on world-wide innovation rates. The same parameters which were shown in Section 2.4 to determine a closed economy’s productivity-growth rate will now determine that country’s relative productivity level. What emerges from this exercise is therefore not just a theory of club convergence but also a theory of the world’s growth rate and of the cross-country distribution of productivity. Before we develop the model we need to address the question of how our framework, in which growth depends on research and development, can be applied to the poorest countries of the world, in which, according to OECD statistics, almost no formal R&D
8 Baumol (1986), Durlauf and Johnson (1995), Quah (1993, 1997) and Mayer-Foulkes (2002, 2003).
78
P. Aghion and P. Howitt
takes place. The key to our answer is that because technological knowledge is often tacit and circumstantially specific,9 foreign technologies cannot simply be copied and transplanted to another country no cost. Instead, technology transfer requires the receiving country to invest resources in order to master foreign technologies and adapt them to the local environment. Although these investments may not fit the conventional definition of R&D, they play the same role as R&D in an innovation-based growth model; that is, they use resources, including skilled labor with valuable alternative uses, they generate new technological possibilities where they are conducted, and they build on previous knowledge.10 While it may be the case that implementing a foreign technology is somewhat easier than inventing an entirely new one, this is a difference in degree, not in kind. In the interest of simplicity our theory ignores that difference in degree and treats the implementation and adaptation activities undertaken by countries far behind the frontier as being analytically the same as the research and development activities undertaken by countries on or near the technological frontier. For all countries we assign to R&D the role that Nelson and Phelps (1966) assumed was played by human capital, namely that of determining the country’s “absorptive capacity”.11 3.1. A model of technology transfer Consider one country in a world of h different countries. This country looks just like the ones described in the basic model above, except that whenever an innovation takes place in any given sector the productivity parameter attached to the new product will be an improvement over the pre-existing global leading-edge technology. That is, let At−1 be the maximum productivity parameter over all countries in the sector at the end of period t − 1; in other words the “frontier” productivity at t − 1. Then an innovation at date t will result in a new version of that intermediate sector whose productivity parameter is At = γ At−1 , which can be implemented by the innovator in this country, and which becomes the new global frontier in that sector. The frontier parameter will also be raised by the factor γ if an innovation occurs in that sector in any other country. Therefore domestic productivity in the sector evolves according to: ln At−1 + ln γ = ln At with probability µ, ln At = with probability 1 − µ ln At−1 where µ is the country’s innovation rate: µ = λf (n) 9 See Arrow (1969) and Evenson and Westphal (1995). 10 Cohen and Levinthal (1989) and Griffith, Redding and Van Reenen (2001) have also argued that R&D by
the receiving country is a necessary input to technology transfer. 11 Grossman and Helpman (1991) and Barro and Sala-i-Martin (1997) also model technology transfer as
taking place through a costly investment process, which they portray as imitation; but in these models technology transfer always leads to convergence in growth rates except in special cases studied by Grossman and Helpman where technology transfer is inactive in the long run.
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
79
and the productivity-adjusted research n is defined as: n ≡ Nt / γ At−1 since the targeted productivity parameter is now γ At−1 . Meanwhile, the global frontier advances by the factor γ with every innovation anywhere in the world.12 Therefore: ln At−1 + ln γ with probability µ, ln At = (11) with probability 1 − µ ln At−1 where
µ = Σ1h λj f nj
is the global innovation rate.13 It follows from (11) that the long-run average growth rate of the frontier, measured as a difference in logs, is:14 g = µ ln γ .
(12) goods.15
Then the Assume there is no international trade in intermediates or general costs and benefits of R&D are the same as in the previous model, except that the domestic productivity parameter At may differ from the global parameter At that research aims to improve upon. Each innovation will now change log productivity by: ln At−1 + ln γ − ln At−1 = ln γ + dt−1 where
dt−1 ≡ ln At−1 /At−1
is a measure of “distance to the frontier”. As Gerschenkron (1952) argued when discussing the “advantage of backwardness”, the greater the distance the larger the innovation. The average growth rate will again be the expected frequency of innovations times 12 Again we are assuming a time period small enough to ignore the possibility of simultaneous innovations
in the same sector. 13 A simpler version of the model would just have the frontier productivity grow at an exogenous rate g. The
model in this section has the advantage of delivering both an endogenous rate for productivity growth at the frontier and club convergence towards that frontier. 14 The growth rate (12) expressed as a log difference is approximately the same as the rate (5) of the previous section which was expressed as a proportional increment, because the first-order Taylor-series approximation to ln γ at γ = 1 is (γ − 1). We switch between these two definitions depending on which is more convenient in a given context. 15 This is not to say that international trade is unimportant for technology transfer. On the contrary, Coe and Helpman (1995), Coe, Helpman and Hoffmaister (1996), Eaton and Kortum (1996) and Savvides and Zachariadis (2005) all provide strong evidence to the effect that international trade plays an important role in the international diffusion of technological progress. For a recent summary of this and other empirical work, see Keller (2002). Eaton and Kortum (2001) provide a simple “semi-endogenous” (see Section 5) growth model in which endogenous innovation interacts with technology transfer and international trade in goods; in their model all countries converge to the same long-run growth rate.
80
P. Aghion and P. Howitt
size: gt = µ(ln γ + dt−1 )
(13)
which is also larger the greater the distance to the frontier. The distance variable dt evolves according to: d with probability 1 − µ, dt =
t−1
ln γ + dt−1 0
with probability µ − µ, with probability µ.
That is, with probability 1 − µ there is no innovation in the sector either globally or in this country, so both domestic productivity and frontier productivity remain unchanged; with probability µ − µ an innovation will occur in this sector but in some other country, in which case domestic productivity remains the same but the proportional gap grows by the factor γ ; and with probability µ an innovation will occur in this sector in this country, in which case the country moves up to the frontier, reducing the gap to zero. It follows that the expected distance d t evolves according to: d t = (1 − µ)d t−1 + (µ − µ) ln γ . If µ > 0 this is a stable difference equation with a unique rest point. That is, as long as the country continues to perform R&D at a positive constant intensity n its distance to the frontier will stabilize, meaning that its productivity growth rate will converge to that of the global frontier. But if µ = 0 the difference equation has no stable rest point and d t diverges to infinity. That is, if the country stops innovating it will have a longrun productivity growth rate of zero because innovation is a necessary condition for the country to benefit from technology transfer. More formally, the country’s long-run expected distance d ∗ is given by: (µ/µ − 1) ln γ if µ > 0, ∗ d = (14) ∞ if µ = 0 and its long-run expected growth rate g ∗ , according to (12), (13) and (14) is: µ(ln γ + d ∗ ) = g if µ > 0, ∗ g = 0 if µ = 0. Each country’s innovation rate µ is determined according to the same principles as before. In particular, it will be equal λf (n) where n is determined by the researcharbitrage equation (9) above, provided that a positive solution to (9) exists. For example if the research-productivity function f satisfies the Inada-like condition: f (0) = ∞, as in the example used above to derive the growth equation (10), then there will always exist a positive solution to (9), so all countries will converge to the frontier growth rate. But suppose, on the contrary, that this Inada-like condition does not hold, that instead: f (0) < ∞. Then the research-arbitrage condition (9) must be replaced by the more general Kuhn–Tucker conditions: 1 λf (n)δ(χ)L/m
with n = 0 if the inequality is strict.
(15)
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
81
That is, for an interior solution the expected marginal cost and benefit must be equal, but the only equilibrium will be one with zero R&D if at that point the expected marginal benefit does not exceed the cost. It follows that the country will perform positive R&D if: λδ(χ)L/m > 1/f (0),
(16)
but if condition (16) fails then there will be no research: n = 0 and hence no innovations: µ = 0 and no growth: g = 0. This means that countries will fall into two groups, corresponding to two convergence clubs: 1. Countries with highly productive R&D, as measured by λ, or good educational systems as measured by high λ or high L, or good property right protection as measured by a high χ, will satisfy condition (16), and hence will grow asymptotically at the frontier growth rate g. 2. Countries with low R&D productivity, poor educational systems and low property right protection will fail condition (16) and will not grow at all. The gap dt separating them from the frontier will grow forever at the rate g. 3.2. World growth and distribution Since the world growth rate g given by (12) depends on each country’s innovation frequency µj = λj f (nj ), therefore world growth depends on the value for each country of all the factors described in Section 2.4 that determine µj . Thus any improvement in R&D productivity, education or property rights anywhere in the innovating world will raise the growth rate of productivity in all but the stagnating countries. Moreover, the cross-country distribution of productivity is determined by these same variables. For according to (14) each country’s long-run relative distance to the frontier depends uniquely on its own innovation frequency µ = λf (n). Two countries in which the determinants of innovation analyzed in Section 2.4 are the same will lie the same distance from the frontier in the long run and hence will have the same productivity in the long run. Countries with more productive R&D, better educational systems and stronger property right protection will have higher productivity. 3.3. The role of financial development in convergence The framework can be further developed by assuming that while the size of innovations increases with the distance to the technological frontier (due to technology transfer), the frequency of innovations depends upon the ratio between the distance to the technological frontier and the current stock of skilled workers. This enriched framework [see Howitt and Mayer-Foulkes (2002)] can explain not only why some countries converge while other countries stagnate but also why different countries may display positive yet divergent growth patterns in the long-run. Benhabib and Spiegel (2002) develop a similar account of divergence and show the importance of human capital in the process.
82
P. Aghion and P. Howitt
The rest of this section presents a summary of the related model of Aghion, Howitt, and Mayer-Foulkes (2004) (AHM) and discusses their empirical results showing the importance of financial development in the convergence process. Suppose that the world is as portrayed in the previous sections, but that research aimed at making an innovation in t must be done at period t − 1. If we assume perfectly functioning financial markets then nothing much happens to the model except that the research arbitrage condition (9) has a discount factor β on the right-hand side to reflect the fact that the expected returns to R&D occur one period later than the expenditure.16 But when credit markets are imperfect, AHM show that an entrepreneur may face a borrowing constraint that limits her investment to a fixed multiple of her accumulated net wealth. In their model the multiple comes from the possibility that the borrower can, at a cost that is proportional to the size of her investment, decide to defraud her creditors by making arrangements to hide the proceeds of the R&D project in the event of success.17 They also assume a two-period overlapping-generations structure in which the accumulated net wealth of an entrepreneur is her current wage income, and in which there is just one entrepreneur per sector in each country. This means that the further behind the frontier the country falls the less will any entrepreneur be able to invest in R&D relative to what is needed to maintain any given frequency of innovation. What happens in the long run to the country’s growth rate depends upon the interaction between this disadvantage of backwardness, which reduces the frequency of innovations, and the above-described advantage of backwardness, which increases the size of innovations. The lower the cost of defrauding a creditor the more likely it is that the disadvantage of backwardness will be the dominant force, preventing the country from converging to the frontier growth rate even in the long run. Generally speaking, the greater the degree of financial development of a country the more effective are the institutions and laws that make it difficult to defraud a creditor. Hence the link between financial development and the likelihood that a country will converge to the frontier growth rate. The following simplified account of AHM shows in more detail how this link between financial development and convergence works. Suppose that entrepreneurs have no source of income other than what they can earn from innovating. Then they must borrow the entire cost of any R&D project. Because there are constant returns to the R&D technology,18 therefore in equilibrium that cost will equal the expected benefit, discounted back to today: Nt = µβπt+1 . This is also the expected discounted benefit to a borrower from paying a cost cNt today that would enable her to default in the event that the R&D project is successful. (There 16 For simplicity we suppose that everyone has linear intertemporal preferences with a constant discount
factor β. 17 The “credit multiple” assumed here is much like that of Bernanke and Gertler (1989), as modified by
Aghion, Banerjee and Piketty (1999). 18 See footnote 2 above.
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
83
is no benefit if the project fails to produce an innovation because in that case the entrepreneur cannot pay anything to the creditor even if she has decided to be honest and therefore has not paid the cost cNt .) The entrepreneur will choose to he honest if the cost at least as great as the benefit; that is, if: c 1.
(17)
Otherwise she will default on any loan. Suppose that βλδ(χ)L/m > 1/f (0). This is the condition (16) above for positive growth, modified to take discounting into account. It follows that in any country where the incentive-compatibility constraint (17) holds then innovation will proceed as described in the previous section, and the country will converge to the frontier growth rate. But in any country where the cost of defrauding a creditor is less than unity no R&D will take place because creditors would rationally expect to be defrauded of any possible return from lending to an entrepreneur. Therefore convergence to the frontier growth rate will occur only in countries with a level of financial development that is high enough to put the cost of fraud at or above the limit of unity imposed by (17). AHM test this effect of financial development on convergence by running the following cross-country growth regression: gi − g1 = β0 + βf Fi + βy · (yi − y1 ) + βfy · Fi · (yi − y1 ) + βx Xi + εi
(18)
where gi denotes the average growth rate of per-capita GDP in country i over the period 1960–1995, Fi the country’s average level of financial development, yi the initial (1960) log of per-capita GDP, Xi a set of other regressors and εi a disturbance term with mean zero. Country 1 is the technology leader, which they take to be the United States. Define yi ≡ yi − y1 , country i’s initial relative per-capita GDP. Under the assumption that βy + βfy Fi = 0 we can rewrite (18) as: gi − g1 = λ i · yi − yi∗ where the steady-state value yi∗ is defined by setting the right-hand side of (18) to zero: yi∗ = −
β0 + βf Fi + βx Xi + εi βy + βfy Fi
(19)
and λi is a country-specific convergence parameter: λi = βy + βfy Fi
(20)
that depends on financial development. A country can converge to the frontier growth rate if and only if the growth rate of its relative per-capita GDP depends negatively on the initial value yi ; that is if and only if the convergence parameter λi is negative. Thus the likelihood of convergence will increase with financial development, as implied by the above theory, if and only if: βfy < 0.
(21)
84
P. Aghion and P. Howitt
The results of running this regression using a sample of 71 countries are shown in Table 1, which indicates that the interaction coefficient βfy is indeed significantly negative for a variety of different measures of financial development and a variety of different conditioning sets X. The estimation is by instrumental variables, using a country’s legal origins,19 and its legal origins interacted with the initial GDP gap (yi − y1 ) as instruments for Fi and Fi (yi − y1 ). The data, estimation methods and choice of conditioning sets X are all take directly from Levine, Loayza and Beck (2000) who found a strongly positive and robust effect of financial intermediation on short-run growth in a regression identical to (18) but without the crucial interaction term Fi (yi − y1 ) that allows convergence to depend upon the level of financial development. AHM shown that the results of Table 1 are surprisingly robust to different estimation techniques, to discarding outliers, and to including possible interaction effects between the initial GDP gap and other right-hand-side variables. 3.4. Concluding remark Thus we see how Schumpeterian growth theory and the quality improvement model can naturally explain club convergence patterns, the so-called twin peaks pointed out by Quah (1996). The Schumpeterian growth framework can deliver an explanation for cross-country differences in growth rates and/or in convergence patterns based upon institutional considerations. No one can deny that such considerations are close to what development economists have been concerned with. However, some may argue that the quality improvement paradigm, and new growth theories in general, remain of little help for development policy, that they merely formalize platitudes regarding the growth-enhancing nature of good property right protection, sound education systems, stable macroeconomy, without regard to specifics such as a country’s current stage of development. In Sections 4 and 6 we will argue on the contrary that the Schumpeterian growth paradigm can be used to understand (i) why liberalization policies (in particular an increase in product market competition) should affect productivity growth differently in sectors or countries at different stages of technological development as measured by the distance variable d; and (ii) why the organizations or institutions that maximize growth, or that are actually chosen by societies, also vary with distance to the frontier.
4. Linking growth to IO: innovate to escape competition One particularly unappealing feature of the basic Schumpeterian model outlined in Section 2 is the prediction that product market competition is unambiguously detrimental to growth because it reduces the monopoly rents that reward successful innovators and
19 See La Porta et al. (1998) for a detailed explanation of legal origins and its relevance as an instrument for
financial development.
Financial Development (F ) Conditioning set (X) Coefficient estimates βf βy βfy Sample size
Private Credit
Liquid Liabilities
Bank Assets
Empty
Policya
Fullb
Empty
Policya
Fulla
Empty
Policya
Fullb
−0.015 (−0.93) 1.507∗∗∗ (3.14) −0.061∗∗∗ (−5.35) 71
−0.013 (−0.68) 1.193∗ (1.86) −0.063∗∗∗ (−5.10) 63
−0.016 (−0.78) 1.131 (1.49) −0.063∗∗∗ (−4.62) 63
−0.029 (−1.04) 2.648∗∗∗ (3.12) −0.076∗∗∗ (−3.68) 71
−0.030 (−0.99) 2.388∗∗ (2.39) −0.077∗∗∗ (−3.81) 63
−0.027 (−0.90) 2.384∗∗ (2.11) −0.073∗∗∗ (−3.55) 63
−0.019 (−1.07) 1.891∗∗∗ (3.57) −0.081∗∗∗ (−5.07) 71
−0.020 (−1.03) 1.335∗ (1.93) −0.081∗∗∗ (−4.85) 63
−0.022 (−1.12) 1.365 (1.66) −0.081∗∗∗ (−4.46) 63
Notes: The dependent variable g − g1 is the average growth rate of per-capita real GDP relative to the US, 1960–95. F is average Financial Development 1960–95 using 3 alternative measures: Private Credit is the value of credits by financial intermediaries to the private sector, divided by GDP, Liquid Liabilities is currency plus demand and interest-bearing liabilities of banks and non-bank financial intermediaries, divided by GDP, and Bank Assets is the ratio of all credits by banks to GDP. y − y1 is the log of per-capita GDP in 1960 relative to the United States. a The Policy conditioning set includes average years of schooling in 1960, government size, inflation, the black market premium and openness to trade.
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
Table 1 Growth, financial development, and initial GDP gap. Estimation of equation: g − g1 = β0 + βf F + βy (y − y1 ) + βfy F (y − y1 ) + βx X
b The Full conditioning set includes the policy set plus indicators of revolutions and coups, political assassinations and ethnic diversity.
Estimation is by IV using L (legal origins) and L(y − y1 ) as instruments for F and F (y − y1 ). The numbers in parentheses are t-statistics. Significance at the 1%, 5% and 10% levels is denoted by *** , ** and * , respectively.
85
86
P. Aghion and P. Howitt
thereby discourages R&D investments. Not only does this prediction contradict a common wisdom that goes back to Adam Smith, but it has also been shown to be (partly) counterfactual [e.g., by Geroski (1995), Nickell (1996), and Blundell et al. (1999)].20 However, as we argue in this section, a simple modification reconciles the Schumpeterian paradigm with the evidence on product market competition and innovation, and also generates new empirical predictions that can be tested with firm- and industrylevel data. In this respect the paradigm can meet the challenge of seriously putting IO into growth theory. The theory developed in this section is based on Aghion, Harris and Vickers (1997) and Aghion et al. (2001), but cast in the discrete-time framework introduced above. 4.1. The theory We start by considering an isolated country in a variant of the technology-transfer model of the previous section. This variant allows technology spillovers to occur across sectors as well as across national borders. Thus there is a global technological frontier that is common to all sectors, and which is drawn on by all innovations. The model takes as given the growth rate of this global frontier, so that the frontier At at the end of period t obeys: At = γ At−1 , where γ > 1. In each country, the general good is produced using the same kind of technology as in the previous sections, but here for simplicity we assume a continuum of intermediate inputs and we normalize the labor supply at L = 1, so that:
1 yt = Ait1−α xitα di, 0
where, in each sector i, only one firm produces intermediate input i using the general good as capital according to a one-for-one technology. In each sector, the incumbent firm faces a competitive fringe of firms that can produce the same kind of intermediate good, although at a higher unit cost. More specifically, we assume that at the end of period t, at unit cost χ, where we assume 1 < χ < 1/α < γ χ, a competitive fringe of firms can produce one unit of intermediate input i of a quality equal to min(Ait , At−1 ), where Ait is the productivity level achieved in sector i after innovation has had the opportunity to occur in sector i within period t. In each period t, there are three types of sectors, which we refer to as type-j sectors, with j ∈ {0, 1, 2}. A type-j sector starts up at the beginning of period t with productivity Aj,t−1 = At−1−j , that is, j steps behind the current frontier At−1 . The profit flow of an 20 We refer the reader to the second part of this section where we confront theory and empirics on the rela-
tionship between competition/entry and innovation/productivity growth.
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
87
incumbent firm in any sector at the end of period t, will depend upon the technological position of that firm with regard to the technological frontier at the end of the period. Between the beginning and the end of the current period t, the incumbent firm in any sector i has the possibility of innovating with positive probability. Innovations occur step-by-step: in any sector an innovation moves productivity upward by the same factor γ . Incumbent firms can affect the probability of an innovation by investing more in R&D at the beginning of the period. Namely, by investing the quadratic R&D effort 1 2 2 γ Ai,t−1 µ an incumbent firm i in a type-0 or type-1 sector innovates with probability µ.21 However, innovation is assumed to be automatic in type-2 sectors, which in turn reflects a knowledge externality from more advanced sectors which limits the maximum distance of any sector to the technological frontier. Now, consider the R&D incentives of incumbent firms in the different types of sectors at the beginning of period t. Firms in type-2 sectors have no incentive to invest in R&D since innovation is automatic in such sectors. Thus µ2 = 0, where µj is the equilibrium R&D choice in sector j . Firms in type-1 sectors, that start one step behind the current frontier at Ai,t−1 = At−2 at the beginning of period t, end up with productivity At = At−1 if they successfully innovate, and with productivity At = At−2 otherwise. In either case, the competitive fringe can produce intermediate goods of the same quality but at cost χ instead of 1, which in turn, as in Section 2, the equilibrium profit is equal to:22 πt = At δ(χ), with 1
δ(χ) = (χ − 1)(χ/α) α−1 . Thus the net rent from innovating for a type-1 firm is equal to At−1 − At−2 δ(χ) and therefore a type-1 firm will choose its R&D effort to solve: 1 2 max At−1 − At−2 δ(χ)µ − γ At−2 µ , µ 2
21 We thus depart slightly from our formulation in the previous sections: here we take the probability of
innovation, not the R&D effort, as the optimization variable. However the two formulations are equivalent: that the innovation probability f (n) = µ is a concave function of the effort n, is equivalent to saying that the effort is a convex function of the probability. 22 Imitation does not destroy the rents of non-innovating firms. We assume nevertheless that the firm ignores any continuation value in its R&D decision.
88
P. Aghion and P. Howitt
which yields 1 δ(χ). µ1 = 1 − γ In particular an increase in product market competition, measured as an reduction in the unit cost χ of the competitive fringe, will reduce the innovation incentives of a type-1 firm. This we refer to as the Schumpeterian effect of product market competition: competition reduces innovation incentives and therefore productivity growth by reducing the rents from innovations of type-1 firms that start below the technological frontier. This is the dominant effect, both in IO models of product differentiation and entry, and in basic endogenous growth models as the one analyzed in the previous sections. Note that type-1 firms cannot escape the fringe by innovating: whether they innovate or not, these firms face competitors that can produce the same quality as theirs at cost χ. As we shall now see, things become different in the case of type-0 firms. Firms in type-0 sectors, that start at the current frontier, end up with productivity At if they innovate, and stay with their initial productivity At−1 if they do not. But the competitive fringe can never get beyond producing quality At−1 . Thus, by innovating, a type-0 incumbent firm produces an intermediate good which is γ times better than the competing good the fringe could produce, and at unit cost 1 instead of χ for the fringe. Our assumption α1 < γ χ then implies that competition by the fringe is no longer a binding constraint for an innovating incumbent, so that its equilibrium profit postinnovation, will simply be the profit of an unconstrained monopolist, namely: πt = At δ(1/α). On the other hand, a type-0 firm that does not innovate will keep its productivity equal to At−1 . Since the competitive fringe can produce up to this quality level at cost χ, the equilibrium profit of a type-0 firm that does not innovate is equal to πt = At−1 δ(χ). A type-0 firm will then choose its R&D effort to:
1 max At δ(1/α) − At−1 δ(χ) µ − γ At−1 µ2 , µ 2 so that in equilibrium 1 δ(χ). γ In particular an increase in product market competition, i.e. a reduction in χ, will now have a fostering effect on R&D and innovation. This, we refer to as the escape competition effect: competition reduces pre-innovation rents of type-0 incumbent firms, but not their post-innovation rents since by innovating these firms have escaped the fringe. This in turn induces those firms to innovate in order to escape competition with the fringe. The combination of these two effects can explain the non-monotonicity of the relationship between competition and growth that we find in the data. µ0 = δ(1/α) −
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
89
4.2. Empirical predictions The above analysis generates interesting predictions: 1. Innovation in sectors in which firms are close to the technology frontier, react positively to an increase in product market competition; 2. Innovation reacts less positively, or negatively, in sectors in which firms are further below the technological frontier. These predictions have been confronted by Aghion et al. (2002) with UK firm level data on competition and patenting, and we briefly summarize their findings in the next subsection. 4.3. Empirical evidence and relationship to literature Most innovation-based growth models – including the quality improvement model developed in the above two sections – would predict that product market competition is detrimental to growth as it reduces the monopoly rents that reward successful innovators (we refer to this as the Schumpeterian effect of competition). However, an increasing number of empirical studies have cast doubt on this prediction. The empirical IO literature on competition and innovation starts with the pioneering work of Scherer (1967), followed by Cohen and Levin (1989), and more recently by Geroski (1995). All these papers point to a positive correlation between competition and growth. However, competition is often measured by the inverse of market concentration, an indicator which Boone (2000) and others have shown to be problematic: namely, higher competition between firms with different unit costs may actually result in a higher equilibrium market share for the low cost firm! More recently, Nickell (1996) and Blundell et al. (1999) have made further steps by conducting cross-industry analyses over longer time periods and by proposing several alternative measures of competition, in particular the inverse of the Lerner index (defined as the ratio of rents over value added) or by the number of competitors for each firm in the survey. However, none of these studies would uncover the reason(s) why competition can be growth-enhancing or why the Schumpeterian effect does not seem to operate. It is by merging the Schumpeterian growth paradigm with previous patent race models (in which each of two incumbent firms would both, compete on the product market and innovate to acquire a lead over its competitor), that Aghion, Harris and Vickers (1997), henceforth AHV, and Aghion et al. (2001), henceforth AHHV, have developed new models of competition and growth with step-by-step innovations that reconcile theory and evidence on the effects of competition and growth: by introducing the possibility that innovations be made by incumbent firms that compete “neck-and-neck”, these extensions of the Schumpeterian growth framework show the existence of an “escape competition” effect that counteracts the Schumpeterian effect described above. What facilitated this merger between the Schumpeterian growth approach and the patent race models, is that: (i) both featured quality-improving innovations; (ii) models with vertical innovations in turn were particularly convenient to formalize the notion of technological
90
P. Aghion and P. Howitt
distance and that of “neck-and-neck” competition. A main prediction of this new vintage of endogenous growth models, is that competition should be most growth-enhancing in sectors in which incumbent firms are close to the technological frontier and/or compete “neck-and-neck” with one another, as it is in those sectors that the “escape competition” effect should be the strongest. These models in turn have provided a new pair of glasses for deeper empirical analyses of the relationship between competition/entry and innovation/growth. The two studies we briefly mention in the remaining part of this section have not only produce interesting new findings; they also suggested a whole new way of confronting endogenous growth theories with data, one that is more directly grounded on serious microeconometric analyses based on detailed firm/industry panels. The paper by Aghion et al. (2002), henceforth ABBGH, takes a new look at the effects of product market competition on innovation, by confronting the main predictions of the AHV and AHHV models to firm level data. The prediction we want to emphasize here as it is very much in tune with our theoretical discussion in the previous subsections, is that the escape competition effect should be strongest in industries in which firms are closest to the technological frontier. ABBGH considers a UK panel of individual companies during the period 1968– 1997. This panel includes all companies quoted on the London Stock Exchange over that period, and whose names begin with a letter from A to L. To compute competition measures, the study uses firm level accounting data from Datastream; product market competition is in turn measured by one minus the Lerner index (ratio of operating profits minus financial costs over sales), controlling for capital depreciation, advertising expenditures, and firm size. Furthermore, to control for the possibility that variations in the Lerner index be mostly due to variations in fixed costs, we use policy instruments such as the implementation of the Single Market Program or lagged values of the Lerner index as instrumental variables. Innovation activities, in turn, are measured both, by the number of patents weighted by citations, and by R&D spending. Patenting information comes from the U.S. Patent Office where most firms that engage in international trade register their patents; in particular, this includes 461 companies on the London Stock Exchange with names starting by A to L, for which we already had detailed accounting data. Finally, technological frontier is measured as follows: suppose a UK firm (call it i) belongs to some industry A; then we measure technological distance by the difference between the maximum TFP in industry A across all OECD countries (we call it TFPF , where the subscript “F ” refers to the technological frontier) and the TFP of the UK firm, divided by the former: mi =
TFPF − TFPi . TFPF
Figure 1 summarizes our main findings. Each point on this figure corresponds to one firm in a given year. The upper curve considers only those firms in industries where the average distance to the technological frontier is less than the median distance across all industries, whereas the lower curve includes firms in all industries.
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
91
Figure 1. Inovation and competention: The neck and neck split. The figure plots a measure of competition on the x-axis again citation weighted patents on the y-axis. Each point represents an industry-year. The circles show the exponential quadratic curve that is reported in column (2) of Table 1. The triangles show the exponential quadratic curve estimated only on neck-and-neck industires that is reported in column (4) of Table 1.
We clearly see that the effect of product market competition on innovation is all the more positive that firms are closer to the technological frontier (or equivalently are more “neck-and-neck”). Another interesting finding that comes out of the figure, is that the Schumpeterian effect is also at work, and that it dominates at high initial levels of product market competition. This in turn reflects the “composition effect” pointed out in the previous subsection: namely, as competition increases and neck-and-neck firms therefore engage in more intense innovation to escape competition, the equilibrium fraction of neck-and-neck industries tends to decrease (equivalently, any individual firm spends less time in neck-and-neck competition with its main rivals) and therefore the average impact of the escape competition effect decreases at the expense of the counteracting Schumpeterian effect. The ABBGH paper indeed shows that the average distance to the technological distance increases with the degree of product market competition. The Schumpeterian effect was missed by previous empirical studies, mainly as a result of their being confined to linear estimations. Instead, more in line with the Poisson technology that governs the arrival of innovations both, in Schumpeterian and in patent race models, ABBGH use a semi-parametric estimation method in which the expected flow of innovations is a piecewise polynomial function of the Lerner index.
92
P. Aghion and P. Howitt
4.4. A remark on inequality and growth Our discussion of the effects of competition on growth also sheds light on the current debate on the effects of income or wealth inequality on growth. A recent literature23 has emphasized the idea that in an economy with credit-constraints, where the poor do not have full access to efficient investment opportunities, redistribution may enhance investment by the poor more than it reduces incentives for the rich, thereby resulting in higher aggregate productive efficiency in steady-state and higher rate of capital accumulation on the transition path to the steady-state. Our discussion of the effects of competition on innovation and growth hints at yet another negative effect of excessive wealth concentration on growth: to the extent that innovative activities tend to be more intense in sectors in which firms or individuals compete “neck-and-neck”, taxing further capital gains by firms that are already well ahead of their rivals in the same sector may enhance the aggregate rate of innovation by shifting the overall distribution of technological gaps in the economy towards a higher fraction of neck-and-neck sectors in steady-state. More generally, having too many sectors in which technological knowledge and/or wealth are highly concentrated, may inhibit growth as it both, discourages laggard firms or potential entrants, and reduces the leader’s incentives to innovate in order to escape competition given that the competitive threat coming from laggards or potential entrants is weak; the leader may actually prefer to invest her wealth into entry deterrence activities. These considerations may in turn explain why, following a high growth period during the industrial revolution in the 19th century, growth slowed down at the turn of the 20th century in France or England at the same time wealth distribution became highly concentrated: the high concentration of wealth that resulted from the industrial revolution, turned the innovators of the mid 19th century into entrenched incumbents with the power to protect their dominant position against competition by new potential entrants.24
5. Scale effects25 5.1. Theory Jones (1995) has pointed out that the simple model of the preceding sections whereby increased population leads to increased growth, by raising the size of the market for a successful entrepreneur and by raising the number of potential R&D workers, is not consistent with post-war evidence. In the United States, for example, the number of
23 For example, see Galor and Zeira (1993), Banerjee and Newman (1993), and Aghion and Bolton (1997). 24 See Piketty et al. (2003). 25 This section draws on Ha and Howitt (2004).
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
93
scientists and engineers engaged in R&D has grown by a factor of five since the 1950s with no significant trend increase in productivity growth. This refutes the version of the basic model in which productivity growth is a function of skilled labor applied to R&D (Section 2.3). Likewise, the fact that productivity-adjusted R&D has grown substantially over the same period rejects the version of the model presented in Section 2 above in which productivity growth is a function of productivity-adjusted research. 5.1.1. The Schumpeterian (fully endogenous) solution Schumpeterian theory deals with this problem of the missing scale effect on productivity growth by incorporating Young’s (1998) insight that as an economy grows, proliferation of product varieties reduces the effectiveness of R&D aimed at quality improvement, by causing it to be spread more thinly over a larger number of different sectors.26 When modified this way the theory is consistent with the observed coexistence of stationary TFP growth and rising R&D input, because in a steady state the growth-enhancing effect of rising R&D input is just offset by the deleterious effect of product proliferation. The simplest way to illustrate this modification is to suppose that the number of sectors m is proportional to the size of population L. For simplicity normalize so that m = L.27 Then the growth equation (10) becomes: g = λ2 δ(χ)(γ − 1).
(22)
It follows directly from comparing (22) with (10) that all the comparative-statics propositions of Section 2.4 continue to hold except that now the growth rate is independent of population size. 5.1.2. The semi-endogenous solution Jones (1999) argues that this resolution of the problem is less intuitively appealing than his alternative semi-endogenous theory, built on the idea of diminishing returns to the stock of knowledge in R&D. In this theory sustained growth in R&D input is necessary just to maintain a given rate of productivity growth. Semi-endogenous growth theory has a stark long-run prediction, namely that the long-run rate of productivity growth, and hence the long-run growth rate of per-capita income, depend on the rate of population growth, which ultimately limits the growth rate of R&D labor, to the exclusion of all economic determinants. In Jones’s formulation: g = λf (n)Aφ−1 (γ − 1),
φ<1
26 Variants of this idea have been explored by van de Klundert and Smulders (1997), Peretto (1998),
Dinopoulos and Thompson (1998) and Howitt (1999). 27 Thus, in contrast to Romer (1990) where horizontal innovations drive the growth process, here product
proliferation eliminates scale effects and long-run growth is still ultimately driven by quality-improving innovations.
94
P. Aghion and P. Howitt
where the R&D input n is measured by the number R&D workers in G5 countries. Except for the assumption of diminishing returns (φ < 1) this is equivalent to the original formulation (5). In the special case where f takes a Cobb–Douglas form we have, in continuous time: ˙ g ≡ A/A = λnσ Aφ−1 (γ − 1) so that: g/g ˙ = (1 − φ)(γ gn − g)
(23)
˙ is the growth rate of R&D workers and γ = σ/(1 − φ). where gn = n/n This semi-endogenous model is compatible with the observation of positive trend growth in R&D input, because as long as φ < 1 and the time path of gn is bounded, the differential equation (23) yields a bounded solution for productivity growth. In particular, if gn is constant, or approaches a constant, then g → γ gn . In the long run the growth rate of R&D labor cannot exceed the growth rate η of population, and in a balanced growth equilibrium it will equal η. Likewise, the growth rate of productivity-adjusted R&D expenditure will equal η along a balanced growth path. Hence the radical implication that the long-run growth rate of an economy will equal γ η, independently of what fraction of society’s resources are assigned to knowledge creation. Policies to stimulate R&D will have at most transitory effects on productivity growth and, by extension, on per-capita income growth. 5.2. Evidence These two competing approaches to reconciling R&D-based theory with the observed upward trend in R&D input offer a stark contrast. The Schumpeterian approach with product-proliferation effects retains all the characteristic comparative statics predictions of endogenous growth theory as outlined in Section 2.4, while Jones’s semi-endogenous theory denies all these predictions. Fortunately the two competing approaches can also be tested using observed trends in productivity growth and R&D input. Specifically, the semi-endogenous model implies that the growth rate of productivity will track the growth rate of R&D input, whereas the Schumpeterian model implies that it will track the fraction of GDP spent on R&D.28 To derive this Schumpeterian implication note that, according to the growth equation (5), productivity growth depends on productivity-adjusted R&D per sector, n. Given the assumption m = L, if GDP per person grows asymptotically at the rate g then n will be proportional to the fraction of GDP spent on R&D. 28 Zachariadis (2003) shows that the fully-endogenous Schumpeterian theory without scale effects also passes
a number of other tests using U.S. data. Specifically, he finds using two-digit industry level data that patenting, technological progress and productivity growth all depend upon the ratio of R&D expenditures to output, as implied by the fully endogenous theory.
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
95
Figure 2. TFP growth rates, U.S., 1950–2000.
Figure 2 shows the growth rate of productivity in the United States from 1950 to 2000. There is no discernible trend. An Augmented Dickey–Fuller test rejects a unit root at the 1% significance level, confirming the stationarity of this series. Thus semiendogenous theory implies that the growth rate of R&D input should also be trendless and stationary, whereas Schumpeterian theory implies that the R&D/GDP ratio should be trendless and stationary. 5.2.1. Results Figure 3 shows that growth rates of the number of R&D workers in the G5 countries, N , and U.S. R&D expenditure, R, appear to have a substantial negative trend, having fallen roughly fourfold since the early 1950s. The impression of non-stationarity is supported by an Augmented Dickey–Fuller test, which fails to reject a unit root in gN at the 5% level.
96
P. Aghion and P. Howitt
Figure 3. Growth rates of G5 R&D workers and U.S. R&D expenditures.
These findings are inconsistent with the implications of semi-endogenous growth theory.29 Indeed they undermine the central proposition of semi-endogenous theory, because if productivity growth can be sustained for 50 years in the face of such a large fall in the growth rate of R&D labor then there is no reason to suppose that population growth limits productivity growth, except perhaps over a time scale of hundreds of years. Figure 4 shows that the fraction of GDP spend on R&D in the U.S. looks more or less stable with perhaps a small upward trend.30 It is notable that ever since 1957, R&D as a percentage of GDP has been fluctuating between 2.1% and 2.9%, with similar 29 The data on G5 R&D workers come from Jones, who had to guess at the non-U.S. component from 1950
to 1965. However, Ha and Howitt (2004) consider a broader range of R&D measures. They also show that the formal cointegration predictions implied by semi-endogenous theory are not found in these data, even if attention is restricted to the post-1965 date, while the even tighter cointegration predictions implied by Schumpeterian theory are found. Ha and Howitt also conduct a calibration exercise and show that the semiendogenous model when fit to pre-2000 U.S. data predicts productivity worse than the fully endogenous model. 30 There appears to be a more significant upward trend if we omit space and defense R&D, as is done by many researchers in the productivity literature on the grounds that they do not find spillovers from these components
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
97
Figure 4. R&D intensity, U.S., 1953–2000.
movements as in productivity growth: downward trend for 1964–1975 and upward trend for 1975–2000. The stationarity of this series is confirmed by an Augmented Dickey– Fuller test, which rejects a unit root at the 1% level. This is in conformity with the version of Schumpeterian theory presented above, adjusted to take into account the effects of product proliferation. 5.3. Concluding remarks The scale effect whereby increased population should lead to increased productivity growth clearly refutes a simple interpretation of the model in Section 2, in which L stands for the number of (skilled) individuals. However, we have shown in this section
of R&D. However, this literature has not allowed for the very long lags with which we think federal R&D has its effects. Moreover, throwing out federal R&D would at times amount to throwing out about 70% of the total.
98
P. Aghion and P. Howitt
that even if we stick to this interpretation of L, a simple variant of the Schumpeterian model can be developed, which carries all the same long-run growth implications except for the scale effect. The rival semi-endogenous theory of Jones (1995), which denies endogenous growth in the very long run, is inconsistent with the observation that productivity growth can be sustained through half a century of falling growth in R&D labor. The analogous implication of amended Schumpeterian theory, namely that productivity growth can be sustained as long as society allocates a constant fraction of its resources to research, is consistent with the evidence. Two brief remarks conclude this section. First, there is no evidence pointing to the absence of a scale effect at the world level or in small closed economies. That the stock of educated labor should affect technological convergence and productivity growth worldwide was first pointed out by Nelson and Phelps (1966). Second, if we replace L by LeN , where eN denotes the quality of the labor force as measured for example by the average number of years in schooling (so that more educated countries have more efficiency units of labor), then even if one eliminates scale effects by taking L = m, there will still remain a “level effect” embodied in the eN term, whereby a higher average number of years of education N has a positive effect on growth. In the next section we show that increasing the fraction of highly educated workers and/or increasing the average number of years in schooling, have a positive impact on the rate of productivity growth, but the extent of which depends upon the country’s distance to the world technology frontier: in particular, the closer a country is to the frontier, the higher is the effect of an additional year of higher education on its rate of productivity growth.
6. Linking growth to institutional change 6.1. From Schumpeter to Gerschenkron By linking growth to innovation and entrepreneurship, and innovation incentives in turn to characteristics of the economic environment, new growth theories made it possible to analyze the interplay between growth and the design of policies and institutions. For example, the basic model developed in Section 2 suggested that long-run growth would be best enhanced by a combination of good property right protection (to protect the rents of innovators against imitation), a good education system (to increase the efficiency of R&D activities and/or the supply of skilled manufacturing labor), and a stable macroeconomy to reduce interest rates (and thereby increase the net present value of innovative rents). Our discussion of convergence in Section 3 then suggested that the same policies or institutions would also increase a country’s ability to join the convergence club. However, new growth theories may be criticized by development economists and policy makers, precisely because of the universal nature of the policy recommendations that appear to follow from them: no matter how developed a country or sector currently is, it seems that one should prescribe the same medicines (legal reform to enforce property
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
99
rights, investment climate favorable to entrepreneurship, education, macrostability, etc.) to maximize the growth prospects of that country or sector. Yet, in his essay on Economic Backwardness in Historical Perspective, Gerschenkron (1952) argues that relatively backward economies could more rapidly catch up with more advanced countries by introducing “appropriate institutions” that are growthenhancing at an early stage of development but may cease to be so at a later stage. Thus, countries like Japan or Korea managed to achieve very high growth rates between 1945 up until the 1990s with institutional arrangements involving long-term relationships between firms and banks, the predominance of large conglomerates, and strong government intervention through export promotion and subsidized loans to the enterprise sector, all of which depart from the more market-based and laissez-faire institutional model pioneered and promoted by the U.S. That growth-enhancing institutions or policies might change with a country’s or sector’s distance to the technological frontier, should not come as a total surprise to our readers at this point: in Section 4, we saw that competition could have opposite effects on innovation incentives depending on whether firms were initially closer to or farther below the fringe in the corresponding industry (it would enhance innovations in neckand-neck industries, and discourage it in industries where innovating firms are far below the frontier). The same type of conclusion turns out to hold true when one looks at the interplay between countries’ distance to the world technology frontier and “openness”. Using a cross-country panel of more than 100 countries over the 1960–2000 period, Acemoglu, Aghion and Zilibotti (2002), henceforth AAZ, regress the average growth rate over a five year period on a country’s distance to the U.S. frontier (measured by the ratio of GDP per capita in that country to per capita GDP in the U.S.) at the beginning of the period. Then, splitting the sample of countries in two groups, corresponding respectively to a high and a low openness group according to Frankel–Romer’s (1999) openness indicator, AAZ show that average growth decreases more rapidly as a country approaches the world frontier when openness is low. Thus, while a low degree of openness does not appear to be detrimental to growth in countries far below the world frontier, it becomes increasingly detrimental to growth as the country approaches the frontier. AAZ repeat the same exercise using entry costs to new firms (measured as in Djankov et al. (2002)) instead of openness, and they obtain a similar conclusion, namely that high entry costs are most damaging to growth when a country is close to the world frontier, unlike in countries far below the frontier. In this section, we shall argue that Gerschenkron’s idea of “appropriate institutions” can be easily embedded into our growth framework, in a way that can help substantiate the following claims: 1. Different institutions or policy designs affect productivity growth differently depending upon a country’s or sector’s distance to the technological frontier. 2. A country’s distance to the technological frontier affects the type of organizations we observe in this country (e.g., bank versus market finance, vertical integration versus outsourcing, etc.).
100
P. Aghion and P. Howitt
The remaining part of the section is organized as follows. We first describe the growth equation which AAZ introduce to embed the notion of “appropriate institutions” into the Schumpeterian growth framework. We then focus on the first question about the effects of institution design on productivity growth, by concentrating on the relationship between growth and the organization of education. Finally, we briefly discuss the effects of distance on equilibrium institutions in a concluding subsection. 6.2. A simple model of appropriate institutions Consider the following variant of the multi-country growth model of Section 3. In each country, a unique general good which also serves as numéraire, is produced competitively using a continuum of intermediate inputs according to:
1 1−α yt = (24) At (i) xt (i)α di, 0
where At (i) is the productivity in sector i at time t, xt (i) is the flow of intermediate good i used in general good production again at time t, and α ∈ [0, 1]. As before, ex post each intermediate good producer faces a competitive fringe of imitators that forces her to charge a limit price pt (i) = χ > 1. Consequently, equilibrium monopoly profits (gross of the fixed cost) are simply given by: πt (i) = δAt (i) where δ ≡ (χ − 1)χ − 1−α . We still let
1 At ≡ At (i) di 1
0
denote the average productivity in the country at date t, At the productivity at the world frontier which we assume to grow at the constant rate g from one period to the next, and at = At /At the (inverse) measure of the country’s distance to the technological frontier at date t. The main departure from the convergence model in Section 3, lies in the equation for productivity growth. Suppose that intermediate firms have two ways to generate productivity growth: (a) they can imitate existing world frontier technologies; (b) they can innovate upon the previous local technology. More specifically, we assume: At (i) = ηAt−1 + γ At−1 ,
(25)
where ηAt−1 and γ At−1 refer respectively to the imitation and innovation components of productivity growth. Imitations use the existing frontier technology at the end of period (t − 1), thus they multiply At−1 , whereas innovations build on the knowledge stock of the country, and therefore they multiply At−1 .
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
101
Now dividing both sides of (25) by At , using the fact that At = (1 + g)At−1 , and integrating over all intermediate sectors i, we immediately obtain the following linear relationship between the country’s distance to frontier at at date t and the distance to frontier at−1 at date t − 1: at =
1 (η + γ at−1 ). 1+g
(26)
This equation clearly shows that the relative importance of innovation for productivity growth, increases as: (i) the country moves closer to the world technological frontier, i.e. as at−1 moves closer to 1, whereas imitation is more important when the country is far below the frontier, i.e. when at−1 is close to zero; (ii) a new technological revolution (e.g., the ITC revolution) occurs that increases the importance of innovation, i.e. increases γ . This immediately generates a theory of “appropriate institutions” and growth: suppose that imitation and innovation activities do not require the same institutions. Typically, imitation activities (i.e. η in Equation (26)) will be enhanced by long-term investments within (large) existing firms, which in turn may benefit from long-term bank finance and/or subsidized credit. On the other hand, innovation activities (i.e. γ ) require initiative, risk-taking, and also the selection of good projects and talents and the weeding out of projects that turn out not to be profitable. This in turn calls for more market-based and flexible institutions, in particular for a higher reliance on market finance and speculative monitoring, higher competition and trade liberalization to weed out the bad projects, more flexible labor markets for firms to select the most talented or best matched employees, non-integrated firms to increase initiative and entrepreneurship downstream, etc. It then follows from Equation (26) that the growth-maximizing institutions will evolve as a country moves towards the world technological frontier. Far below the frontier, a country will grow faster if it adopts what AAZ refers to as investment-based institutions or policies, whereas closer to the frontier growth will be maximized if the country switches to innovation-based institutions or policies. A natural question is of course whether institutions actually change when they should from a growth- (or welfare-) maximizing point of view, in other words how do equilibrium institutions at all stages of development compare with the growth-maximizing institutions? This question is addressed in details in AAZ, and we will come back to it briefly in the last subsection. 6.3. Appropriate education systems In his seminal paper on economic development, Lucas (1988) emphasized the accumulation of human capital as a main engine of growth; thus, according to the analysis in that paper, cross-country differences in growth rates across countries are primarily
102
P. Aghion and P. Howitt
attributable to differences in rates of accumulation of human capital. An alternative approach, pioneered by Nelson and Phelps (1966), revived by the Schumpeterian growth literature,31 would instead emphasize the combined effect of the stock of human capital and of the innovation process in generating long-run growth and fostering convergence. In this alternative approach, differences in growth rates across countries are mainly attributable to differences in stocks of human capital, which in turn condition countries’ ability to innovate or adapt to new technologies and thereby catch up with the world technological frontier. Thus, in the basic model of Section 2, the equilibrium R&D investment and therefore the steady-state growth rate were shown to be increasing in the aggregate supply of (skilled) labor L and in the productivity of research λ, both of which refer more to the stock and efficiency of human capital than to its rate of accumulation. Now, whichever approach one takes, and the evidence so far supports the two approaches as being somewhat complementary, one may worry about growth models delivering too general a message, namely that more education is always growth enhancing. In this subsection we try to go one step further and argue that the AAZ specification (summarized by Equation (25)), can be used to analyze the effects, not only of the total amount of education, but more importantly of the organization of education, on growth in countries at different stages of development. This subsection, which is based on Aghion, Meghir and Vandenbussche (2003),32 henceforth AMV, focuses on one particular aspect of the organization of education systems, namely the mix between primary, secondary, and higher education. We consider a variant of the AAZ model outlined in the previous subsection, in which innovation requires highly educated labor, whereas imitation can be performed by both, highly educated and lower-skill workers. A main prediction emerging from this model, is that the closer a country gets to the world technology frontier, the more growth-enhancing it becomes to invest in higher education. In the latter part of the subsection we confront this prediction with preliminary cross-country evidence. 6.3.1. Distance to frontier and the growth impact of higher education There is again a unique general good, produced competitively using a continuum of intermediate inputs according to:
1 y= (27) A(i)1−α x(i)α di, 0
where A(i) is the productivity in sector i, x(i) is the flow of intermediate good i used in general good production, α ∈ [0, 1]. 31 For example, see Acemoglu, Aghion, and Zilibotti (2002), Acemoglu et al. (2003), Aghion, Howitt and
Violante (2002) and Aghion (2002). 32 See also Aghion et al. (2005).
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
103
In each intermediate sector i, one intermediate producer can produce the intermediate good with leading-edge productivity At (i), using general good as capital according to a one-for-one technology. As before, ex post each intermediate good producer faces a competitive fringe of imitators that forces her to charge a limit price p(i) = χ > 1. Consequently, we have: ∂y , ∂x so that equilibrium monopoly profits in each sector i are given by: π(i) = p(i) − 1 x(i) = δπ(i) = δA(i)L p(i) = χ =
−1
where δ = (χ − 1)( χα ) 1−α . As in the previous subsection, intermediate firms can increase productivity, either by imitating frontier technologies or by innovating upon existing technologies in the country. Imitation can be performed by both types of workers, whereas innovation requires high education. More specifically, we focus on the following class of productivity growth functions: φ
1−φ
1−σ At (i) − At−1 (i) = uσm,i,t sm,i,t At−1 + γ un,i,t sn,i,t At−1 ,
(28)
where um,i,t (resp. sm,i,t ) is the amount of unskilled (resp. skilled) labor used in imitation in sector i at time t, un,i,t (resp. sn,i,t ) is the amount of unskilled (resp. skilled) units of labor used by sector i in innovation at time t, σ (resp. φ) is the elasticity of unskilled labor in imitation (resp. innovation), and γ > 0 measures the relative efficiency of innovation compared to imitation in generating productivity growth. We assume: A SSUMPTION 1. The elasticity of skilled labor is higher in innovation than in imitation, and conversely for the elasticity of unskilled labor, that is: φ < σ . Let S (resp. U = 1 − S) denote the fraction of the labor force with higher (resp. primary or secondary) education. Let wu,t At−1 (resp. ws,t At−1 ) denote the current price of unskilled (resp. skilled) labor. The total labor cost of productivity improvement by intermediate firm i at time t, is thus equal to:
Wi,t = wu,t (um,i,t + un,i,t ) + ws,t (sm,i,t + sn,i,t ) At−1 . Letting at = At /At measure the country’s distance to the technological frontier, and letting the frontier technology At grow at constant rate g, the intermediate producer will maximize profits net of total labor costs, namely: σ
φ 1−φ 1−σ max δ um,i,t sm,i,t (1 − at−1 ) + γ un,i,t sn,i,t at−1 At−1 − Wi,t . (29) um,i,t ,sm,i,t ,un,i,t ,sn,i,t
104
P. Aghion and P. Howitt
Using the fact that all intermediate firms face the same maximization problem, and that there is a unit mass of intermediate firms, we necessarily have: uj,i,t ≡ uj,t ;
sj,i,t ≡ sj,t
for all i and for j = m, n;
(30)
and in equilibrium: S = sm,t + sn,t ;
U = 1 − S = um,t + un,t .
(31)
Taking first order conditions for the maximization problem (29), then making use of (30) and (31), and then computing the equilibrium rate of productivity growth
1 At (i) − At−1 di, gt = At−1 0 one can establish [see AMV (2003)]: (1−φ) L EMMA 1. Let ψ = σ(1−σ )φ . If parameter values are such that the solution to (29) is interior, then we have:
∂gt = φ(1 − φ)h (a)h(a)1−φ h(a)U − S , ∂a where 1 S (1 − σ )ψ σ (1 − a) σ −φ . h(a) = (1 − φ)γ a U
This, together with the fact that h(a) is obviously decreasing in a given our Assumption 1, immediately implies: P ROPOSITION 1. A marginal increase in the fraction of labor with higher education, enhances productivity growth all the more the closer the country is from the world technology frontier, that is: ∂ 2 gt > 0. ∂a∂S The intuition follows directly from the Rybczynski theorem in international trade. Stated in the context of a two sector-two input economy, this theorem says that an increase in the supply of input in the sector that uses that input more intensively, should increase “output” in that sector more than proportionally. To transpose this result to the context of our model, consider the effect of an increase in the supply of skilled labor, keeping the supply of unskilled labor fixed and for given a. Given that skilled workers contribute relatively more to productivity growth and profits if employed in innovation rather than in imitation (our Assumption 1), the demand for additional skilled labor will tend to be higher in innovation. But then the marginal productivity of unskilled labor should also increase more in innovation than in imitation, hence a net flow of unskilled
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
105
workers should also move from imitation into innovation. This in turn will enhance further the marginal productivity of skilled labor in innovation, thereby inducing an ever greater fraction of skilled labor to move to innovation. Now the closer the country is to the technology frontier (i.e. the higher a), the stronger this Rybszynski effect as a higher a increases the efficiency of both, skilled and unskilled labor, in innovation relative to imitation. A second, reinforcing, reason is that an increase in the fraction of skilled labor reduces the amount of unskilled labor available in the economy, hence reducing the marginal productivity of skilled labor in imitation, all the more the closer the country is from the frontier. We can now confront this prediction with cross-country evidence on higher education, distance to frontier, and productivity growth. 6.3.2. Empirical evidence The prediction that higher education has stronger growth-enhancing effects close to the technological frontier can be tested using cross-regional or cross-country data. Thus VAM consider a panel dataset of 19 OECD countries over the period 1960–2000. Output and investment data are drawn from Penn World Tables 6.1 (2002) and human capital data from Barro and Lee (2000). The Barro–Lee data indicate the fraction of a country’s population that has reached a certain level of schooling at intervals of five years, so VAM use the fraction that has received some higher education together with their measure of TFP (itself constructed assuming a constant labor share of 0.7 across countries) to perform the following regression: gj,t = α0,j + α1 distj,t−1 + α2 Λj,t−1 + α3 distj,t−1 · Λj,t−1 + uj,t , where gj,t is country j ’s growth rate over a five year period, distj,t−1 is country j ’s closeness to the technological frontier at t − 1 (i.e. 5 years before), Λj,t−1 is the fraction of the working age population with some higher education in the previous period and α0,j is a country dummy controlling for country fixed effects. The closeness variable is instrumented with its lagged value at t − 2, and the fraction variable is instrumented using expenditure on tertiary education per capita lagged by two periods, and the interaction term is instrumented using the interaction between the two instruments for closeness and for the fraction variables. Finally, the standard errors we report allow for serial correlation and heteroskedasticity. The results from this regression are shown in Table 2. In particular, as long as we do not fully control for country fixed effects33 we find a positive and significant interaction between our education measure and a country’s closeness to the frontier, as predicted by the theory in the previous subsection. This result demonstrates that it is more important for growth to expand years of higher education close to the technological frontier. 33 The result is robust to controlling for group fixed effects, having regrouped the countries according to the
coefficients on country dummies in the regression with country fixed effects. It is the scarcity of observations that explains why we lose significance when fully controlling for country fixed effects.
106
P. Aghion and P. Howitt Table 2 TFP growth equation (fractions BL) [1]
[2]
[3]
[4]
[5]
−0.13 (.075) −0.025 (.094) –
−0.216 (.287) 0.65 (.63) –
Country dummies p-value country dummies Proximity threshold
No – –
Yes – –
Rank test (p value) Number of observations
– 122
– 122
−0.27 (.063)∗∗∗ −0.89 (.26)∗∗∗ 1.07 (.28)∗∗∗ No – 0.832 (.044) – 122
−0.24 (.29) 0.3 (1.8) 0.4 (1.6) Yes 0 – – 0.13 122
−0.28 (.08)∗∗∗ −0.43 (.24)∗ 1.11 (.3)∗∗∗ Groups – 0.387 (.14) – 122
Proximity Fraction Proximity · Fraction
Note: standard errors in parentheses. Time dummies not reported. In column [5], countries are grouped in the following way: Group 1: Canada, New Zealand, USA; Group 2: Austria, Ireland, Italy, Norway, Portugal; Group 3: Belgium, Finland, France, United Kingdom; Group 4: Denmark, Netherlands, Spain, Sweden, Switzerland; Group 5: Australia. Proximity threshold indicates the value of Proximity above which Fraction is growth-enhancing. One, two and three * indicate significance at the 10, 5 and 1% level, respectively.
7. Conclusion In this chapter we argued that the endogenous growth model with quality-improving innovations provides a framework for analyzing the determinants of long-run growth and convergence that is versatile, simple and empirically useful. Versatile, as the same framework can be used to analyze how growth interacts with development and crosscountry convergence and divergence, how it interacts with industrial organization and in particular market structure, and how it interacts with organizations and institutional change. Simple, since all these aspects can be analyzed using the same elementary formalization. Empirically useful, as the framework generates a whole range of new microeconomic and macroeconomic predictions and also stands up to empirical criticisms better than other endogenous growth models in the literature. Far from closing the field, the chapter suggests many avenues for future research. For example, on growth and convergence, more research remains to be done to identify the main determinants of cross-country convergence and divergence.34 Also important, is to analyze the role of international intellectual property right protections and foreign 34 In Aghion, Howitt, and Mayer-Foulkes (2004) we emphasize the role of credit constraints in R&D as a
distinguishing factor between the countries that converge in growth rates and in levels towards the frontier, those that converge only in growth rates, and those that follow a divergent path towards a lower rate of long-run growth. Whether credit constraints, or other factors such as health, education, and property rights protection, are key to this three-fold classification, remains an open question.
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
107
direct investment in preventing or favoring convergence. On growth and industrial organization, we have restricted attention to product market competition among existing firms. But what can we say about entry and its impact on incumbents’ innovation activities?35 On institutions, we have just touched upon the question of how technical change interacts with organizational change. Do countries or firms/sectors actually get stuck in institutional traps of the kind described in Section 6? What enables such traps to disappear over time? How do political economy considerations interact with this process? There is also the whole issue of wage inequality and its interplay with technical change, on which the Schumpeterian approach developed in this chapter can also shed light.36 If we just had to select three aspects or questions, so far largely open, and which could also be explored using our approach, we would suggest the following. First, on the role of basic science in generating (very) long-term growth. Do fundamental innovations (or the so called “general purpose technologies”) require the same incentive system and the same rewards as industrial innovations? How can one design incentive systems in universities so that university research would best complement private research? A second aspect is the interplay between growth and volatility. Is R&D and innovation procyclical or countercyclical, and is macroeconomic volatility always detrimental to innovation and growth? Answering this question in turn opens up a whole new research topic on the macropolicy of growth.37 A third aspect is the extent to which our growth paradigm can be applied to less developed economies. In particular, can we use the new growth approach developed in this chapter to revisit the important issue of poverty reduction?38 Finally, in this chapter we have argued that modelling growth as resulting from quality-improving innovations, provides a natural framework to address a whole array of issues from competition to development, each time with theoretical predictions that can be empirically tested and also lead to more precise policy prescriptions. However, one might think of more direct ways of testing the quality-ladder model against the variety model analyzed in the other chapters. For example, in current work with Pol Antras and Susanne Prantl, we are using a panel data set of UK firms over the past fifteen years, to assess whether variety had any impact on innovation and growth. Using input–output tables, our preliminary results suggest that exit of input firms has but a positive effect on the productivity growth of final producers.
References Acemoglu, D., Aghion, P., Griffith, R., Zilibotti, F. (2003). “Technology, hold-up, and vertical integration: What do we learn from micro data?”. Mimeo IFS. London.
35 See Aghion and Griffith (2005). 36 See the chapter by Krusell and Violante in this Handbook volume. 37 See Aghion et al. (2004). 38 See Aghion and Armendariz de Aghion (2004). for some preliminary thoughts on this aspect.
108
P. Aghion and P. Howitt
Acemoglu, D., Aghion, P., Zilibotti, F. (2002). “Distance to Frontier, Selection, and Economic Growth”. NBER Working Paper 9066. Aghion, P., Armendariz de Aghion, B. (2004). “A new growth approach to poverty reduction”. Mimeo. Harvard. Aghion, P., Banerjee, A.V., Piketty, T. (1999). “Dualism and macroeconomic volatility”. Quarterly Journal of Economics 114, 1359–1397. Aghion, P., Bolton, P. (1997). “A model of trickle-down growth and development”. Review of Economic Studies. Aghion, P., Griffith, R. (2005). Competition and Growth. MIT Press, Cambridge, MA. Aghion, P., Harris, C., Vickers, J. (1997). “Competition and growth with step-by-step innovation: An example”. European Economic Review Papers and Proceedings, 771–782. Aghion, P., Howitt, P. (1992). “A model of growth through creative destruction”. Econometrica 60, 323–351. Aghion, P., Howitt, P. (1998). Endogenous Growth Theory. MIT Press, Cambridge, MA. Aghion, P., Howitt, P., Mayer-Foulkes, D. (2004). “The effect of financial development on convergence: Theory and evidence”. Unpublished. Brown University. Aghion, P., Howitt, P., Violante, G.L. (2002). “General purpose technology and wage inequality”. Journal of Economic Growth 7, 315–345. Aghion, P., Meghir, C., Vandenbussche, J. (2003). “Productivity growth and the composition of education spending”. Unpublished. Aghion, P., Harris, C., Howitt, P., Vickers, J. (2001). “Competition, imitation, and growth with step-by-step innovation”. Review of Economic Studies 68, 467–492. Aghion, P., Bloom, B., Blundell, R., Griffith, R., Howitt, P. (2002). “Competition and innovation: An invertedU relationship”. NBER Working Paper 9269. Aghion, P., Angeletos, M., Banerjee, A., Manova, K. (2004). “Volatility and growth: Financial development and the cyclical composition of investment”. Working Paper. Harvard. Aghion, P., Boustan, L., Hoxby, C., Vandenbussche, J. (2005). “Exploiting states’ mistakes to identify the impact of higher education on growth”. Mimeo. Harvard. Arrow, K.J. (1969). “Classificatory notes on the production and transmission of technological knowledge”. American Economic Review Papers and Proceedings 59, 29–35. Banerjee, A., Newman, A. (1993). “Occupational choice and the process of development”. Journal of Political Economy 101, 274–298. Barro, R., Lee, J.-W., (2000). “International data and educational attainment: Updates and implications”. Working Paper No 42. Center for Economic Development, Harvard University. Barro, R.J., Sala-i-Martin, X. (1992). “Convergence”. Journal of Political Economy 100, 223–251. Barro, R.J., Sala-i-Martin, X. (1997). “Technological diffusion, convergence, and growth”. Journal of Economic Growth 2, 1–26. Baumol, W.J. (1986). “Productivity growth, convergence, and welfare”. American Economic Review 76, 1072–1085. Benhabib, J., Spiegel, M. (2002). “Human capital and technology diffusion”. Unpublished. NYU. Bernanke, B., Gertler, M. (1989). “Agency costs, net worth, and business fluctuations”. American Economic Review 79, 14–31. Blundell, R., Griffith, R., Van Reenen, J. (1999). “Market share, market value and innovation in a panel of British manufacturing firms”. Review of Economic Studies 66, 529–554. Boone, J., (2000). “Measuring product market competition”. CEPR Discussion Paper 2636. Coe, D.T., Helpman, E. (1995). “International R&D spillovers”. European Economic Review 39, 859–887. Coe, D.T., Helpman, E., Hoffmaister, A.W. (1996). “North–South R&D spillovers”. Economic Journal 107, 134–149. Cohen, W., Levin, R. (1989). “Empirical studies of innovation and market structure”. In: Schalensee, R., Willig, R. (Eds.), Handbook of Industrial Organization. Elsevier, Amsterdam. Chapter 18. Cohen, W.M., Levinthal, D.A. (1989). “Innovation and learning: The two faces of R&D”. Economic Journal 99, 569–596.
Ch. 2: Growth with Quality-Improving Innovations: An Integrated Framework
109
Corriveau, L. (1991). “Entrepreneurs, growth and cycles”. Ph.D. Dissertation. University of Western Ontario. Dinopoulos, E., Thompson, P. (1998). “Schumpeterian growth without scale effects”. Journal of Economic Growth 3, 313–335. Djankov, S., La Porta, R., Lopez-de-Silanes, F., Shleifer, A. (2002). “The regulation of entry”. Quarterly Journal of Economics, CXVII. Durlauf, S., Johnson, P. (1995). “Multiple regimes and cross-country growth behavior”. Journal of Applied Econometrics 10, 365–384. Eaton, J., Kortum, S. (1996). “Trade in ideas: Patenting and productivity in the OECD”. Journal of International Economics 40, 251–278. Eaton, J., Kortum, S. (2001). “Technology, trade, and growth: A unified framework”. European Economic Review 45, 742–755. Evans, P. (1996). “Using cross-country variances to evaluate growth theories”. Journal of Economic Dynamics and Control 20, 1027–1049. Evenson, R.E., Westphal, L.E. (1995). “Technological change and technology strategy”. In: Srinivasan, T.N., Behrman, J. (Eds.), In: Handbook of Development Economics, vol. 3. Elsevier, Amsterdam, pp. 2209– 2299. Frankel, J., Romer, D. (1999). “Does trade cause growth?”. American Economic Review 89, 379–399. Galor, O., Zeira, J. (1993). “Income distribution and macroeconomics”. Review of Economic Studies 60, 35–52. Geroski, P. (1995). Market Structure, Corporate Performance and Innovative Activity. Oxford University Press. Gerschenkron, A. (1952). “Economic backwardness in historical perspective”. In: Hoselitz, B.F. (Ed.), The Progress of Underdeveloped Areas. University of Chicago Press, Chicago. Griffith, R., Redding, S., Van Reenen, J. (2001). “Mapping the two faces of R&D: Productivity growth in a panel of OECD industries”. Unpublished. Grossman, G., Helpman, E. (1991). Innovation and Growth in the Global Economy. MIT Press, Cambridge, MA. Ha, J., Howitt, P. (2004). “Accounting for trends in productivity and R&D: A schumpeterian critique of semiendogenous growth theory”. Unpublished. Howitt, P. (1999). “Steady endogenous growth with population and R&D inputs growing”. Journal of Political Economy 107, 715–730. Howitt, P. (2000). “Endogenous growth and cross-country income differences”. American Economic Review 90, 829–846. Howitt, P., Mayer-Foulkes, D. (2002). “R&D, implementation and stagnation: A Schumpeterian theory of convergence clubs”. NBER Working Paper 9104. Jones, C. (1995). “R&D-based models of economic growth”. Journal of Political Economy 103, 759–784. Jones, C. (1999). “Growth: With or without scale effects?”. American Economic Review, Papers and Proceedings 89, 139–144. Keller, W. (2002). “Technology diffusion and the world distribution of income: The role of geography, language, and trade”. Unpublished. University of Texas. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W. (1998). “Law and finance”. Journal of Political Economy 106, 1113–1155. Levine, R., Loayza, N., Beck, T. (2000). “Financial intermediation and growth: Causality and causes”. Journal of Monetary Economics 46, 31–77. Lucas, R. (1988). “On the mechanics of economic development”. Journal of Monetary Economics 22, 3–42. Maddison, A. (2001). The World Economy: A Millennial Perspective. Development Centre, Paris. Mankiw, N.G., Romer, D., Weil, D.N. (1992). “A contribution to the empirics of economic growth”. Quarterly Journal of Economics 107, 407–437. Mayer-Foulkes, D., 2002. “Global divergence”. Documento de Trabajo del CIDE. SDTE 250, División de Economía.
110
P. Aghion and P. Howitt
Mayer-Foulkes, D. (2003). “Convergence clubs in cross-country life expectancy dynamics”. In: van der Hoeven, R., Shorrocks, A.F. (Eds.), Perspectives on Growth and Poverty. United Nations University Press, Tokyo, pp. 144–171. Nelson, R., Phelps, E. (1966). “Investment in humans, technological diffusion, and economic growth”. American Economic Review 61, 69–75. Nickell, S. (1996). “Competition and corporate performance”. Journal of Political Economy 104, 724–746. Peretto, P.F. (1998). “Technological change, market rivalry, and the evolution of the capitalist engine of growth”. Journal of Economic Growth 3, 53–80. Piketty, T., Postel-Vinay, G., Rosenthal, J.-L. (2003). “Wealth concentration in a developing economy: Paris and France, 1807–1994”. Mimeo. EHESS (Paris). Pritchett, L. (1997). “Divergence, big-time”. Journal of Economic Perspectives 11, 3–17. Quah, D. (1993). “Empirical cross-section dynamics in economic growth”. European Economic Review 37, 426–434. Quah, D. (1996). “Convergence empirics across economies with (some) capital mobility”. Journal of Economic Growth 1, 95–124. Quah, D. (1997). “Empirics for growth and distribution: Stratification, polarization, and convergence clubs”. Journal of Economic Growth 2, 27–59. Romer, P. (1990). “Endogenous technical change”. Journal of Political Economy. Savvides, A. (2005). Zachariadis, M., “International technology diffusion and the growth of TFP in the manufacturing sector of developing economies”. Review of Development Economics 9 (4). Scherer, F. (1967). “Market structure and the employment of scientists and engineers”. American Economic Review 57, 524–531. Segerstrom, P., Anant, T., Dinopoulos, E. (1990). “A Schumpeterian model of the product life cycle”. American Economic Review 80, 1077–1092. Tirole, J. (1988). The Theory of Industrial Organization. MIT Press, Cambridge MA. van de Klundert, T., Smulders, S. (1997). “Growth, competition, and welfare”. Scandinavian Journal of Economics 99, 99–118. Young, A. (1998). “Growth without scale effects”. Journal of Political Economy 106, 41–63. Zachariadis, M. (2003). “R&D, innovation, and technological progress: A test of the Schumpeterian framework without scale effects”. Canadian Journal of Economics 36, 566–586.
Chapter 3
HORIZONTAL INNOVATION IN THE THEORY OF GROWTH AND DEVELOPMENT GINO GANCIA CREI and Universitat Pompeu Fabra FABRIZIO ZILIBOTTI Institute for International Economic Studies, Stockholm University
Contents Abstract Keywords 1. Introduction 2. Growth with expanding variety 2.1. The benchmark model 2.2. Two variations of the benchmark model: “lab-equipment” and “labor-for intermediates” 2.3. Limited patent protection
3. Trade, growth and imitation 3.1. Scale effects, economic integration and trade 3.2. Innovation, imitation and product cycles
4. Directed technical change 4.1. Factor-biased innovation and wage inequality 4.2. Appropriate technology and development 4.3. Trade, inequality and appropriate technology
5. Complementarity in innovation 6. Financial development 7. Endogenous fluctuations 7.1. Deterministic cycles 7.2. Learning and sunspots
8. Conclusions Acknowledgements References
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01003-8
112 112 113 116 116 120 122 124 124 127 130 131 136 140 144 150 157 158 162 166 166 166
112
G. Gancia and F. Zilibotti
Abstract We analyze recent contributions to growth theory based on the model of expanding variety of Romer [Romer, P. (1990). “Endogenous technological change”. Journal of Political Economy 98, 71–102]. In the first part, we present different versions of the benchmark linear model with imperfect competition. These include the “lab-equipment” model, “labor-for-intermediates” and “directed technical change”. We review applications of the expanding variety framework to the analysis of international technology diffusion, trade, cross-country productivity differences, financial development and fluctuations. In many such applications, a key role is played by complementarities in the process of innovation.
Keywords appropriate technology, complementarity, cycles, convergence, directed technical change, endogenous growth, expanding variety, financial development, imperfect competition, integration, innovation, intellectual property rights, imitation, knowledge, learning, patents, technical change, trade, traps JEL classification: D92, E32, F12, F15, F43, G22, O11, O16, O31, O33, O41, O47
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
113
1. Introduction Endogenous growth theory formalizes the role of technical progress in explaining modern economic growth. Although this is a relatively recent development, many of its ideas were already stressed by authors such as Kuznets, Griliches, Schmookler, Rosenberg and Schumpeter. During the 1950s and 1960s, mainstream economics was dominated by the one-sector neoclassical growth model of Solow (1956) and Swan (1956), whose main focus was on capital accumulation. The model postulated the existence of an aggregate production function featuring constant returns to scale and returns to each input falling asymptotically to zero; given that some inputs cannot be accumulated, the model could not generate sustained growth unless technology was assumed to improve exogenously. This simple treatment of technology as exogenous was considered as unsatisfactory for two main reasons: first, by placing the source of sustained growth outside the model, the theory could not explain the determinants of long-run economic performance and second, empirical evidence pointed out that technical progress often depends on deliberate economic decisions. The first attempts to endogenize the rate of technical change addressed the first, but not the second, problem. Assuming technical progress to be an unintentional by-product of the introduction of new capital goods through a process named “learning-by-doing”, Arrow (1962) was able to generate sustained growth at a rate that depended on investment decisions. Attempts at explicitly modeling investment in innovation faced another difficulty. A replication argument suggests that, for a given state of technology, production functions should exhibit constant returns to scale. If technical progress is considered as an additional input, however, the technology features increasing returns to scale and inputs cannot be paid their marginal product. Models of learning-by-doing avoided the problem by assuming that increasing returns were external to firms, thereby preserving perfect competition. However, this approach is not viable once investment in technology is recognized as intentional. The solution was to follow the view of Schumpeter (1942), that new technologies provide market power and that investment in innovation is motivated by the prospect of future profits. In this spirit, Shell (1973) studied the case of a single monopolist investing in technical change and Nordhaus (1969a) wrote a growth model with patents, monopoly power and many firms. In neither case did the equilibrium feature sustained growth.1 A tractable model of imperfect competition under general equilibrium was not available until the analysis of monopolistic competition in consumption goods by Dixit and Stiglitz (1977), later extended to differentiated inputs in production by Ethier (1982). These models also showed how increasing returns could arise from an expansion in the number of varieties of producer and consumer goods, an idea that is at the core of the models studied in this chapter. The first dynamic models of economic growth with monopolistic competition and innovation motivated by profits were built by Judd (1985) 1 See Levhari and Sheshinski (1969) on necessary and sufficient conditions for the existence of steady-state growth in the presence of increasing returns to scale.
114
G. Gancia and F. Zilibotti
and Grossman and Helpman (1989). Yet, these authors were interested in aspects other than endogenous growth and none of their models featured long-run growth. Romer (1987), who formalized an old idea of Young (1928), was the first to show that models of monopolistic competition could generate long-run growth through the increased specialization of labor across an increasing range of activities. The final step was taken in Romer (1990), which assumed that inventing new goods is a deliberate costly activity and that monopoly profits, granted to innovators by patents, motivate discoveries. Since then, the basic model of endogenous growth with an expanding variety of products has been extended in many directions. The distinctive feature of the models discussed in this chapter is “horizontal innovation”: a discovery consists of the technical knowledge required to manufacture a new good that does not displace existing ones. Therefore, innovation takes the form of an expansion in the variety of available products. The underlying assumption is that the availability of more goods, either for final consumption or as intermediate inputs, raises the material well-being of people. This can occur through various channels. Consumers may value variety per se. For example, having a TV set and a Hi-Fi yields more utility than having two units of any one of them. Productivity in manufacturing may increase with the availability of a larger set of intermediate tools, such as hammers, trucks, computers and so on. Similarly, specialization of labor across an increasing variety of activities, as in the celebrated Adam Smith example of the pin factory, can make aggregate production more efficient. The main alternative approach is to model innovation as quality improvements on a given array of products (“vertical innovation”), so that technical progress makes existing products obsolete. This process of “creative destruction” was emphasized by Schumpeter and has been formalized in Aghion and Howitt (1992), Grossman and Helpman (1991a) and Segerstrom, Anant and Dinopoulos (1990). The two approaches naturally complement each other. The main advantage of models with horizontal innovation lies in their analytical tractability, making them powerful tools for addressing a wide range of questions. However, because of their simplistic view on the interaction between innovators, these models are less suited to studying the effects of competition between “leaders” and “followers” on the growth process. Section 1 of this chapter describes a simplified version of Romer (1990) and some extensions used in the literature. The model exhibits increasing returns to scale and steady-state endogenous growth in output per capita and the stock of knowledge. The key feature of the theory is the emphasis on investments in technical knowledge as the determinant of long-run economic growth. Ideas and technological improvements differ from other physical assets, because they entail important public good elements. Inventing new technology is typically costly, while reproducing ideas is relatively inexpensive. Therefore, technical knowledge is described as a non-rival good. Nevertheless, firms are willing to invest in innovation because there exists a system of intellectual property rights (patents) guaranteeing innovators monopoly power over the production and sales of particular goods.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
115
Growth models with an expanding variety of products are a natural dynamic counterpart to trade models based on increasing returns and product differentiation. As such, they offer a simple framework for studying the effects of market integration on growth and other issues in dynamic trade theory. This is the subject of Section 2, which shows how trade integration can produce both static gains, by providing access to foreign varieties, and dynamic gains, by raising the rate at which new goods are introduced. Product-cycle trade and imitation are also considered. In many instances, technical progress may be non-neutral towards different factors or sectors. This possibility is considered in Section 3, where biased technical change is incorporated in the basic growth model. By introducing several factors and sectors, the economic incentives to develop technologies complementing a specific factor, such as skilled workers, can be studied. These incentives critically depend on the definition of property rights over the production of new ideas. The high variability in the effectiveness of patent laws across countries has important bearings on the form of technical progress. In particular, governments in less developed countries may have an incentive not to enforce intellectual property rights in order to speed up the process of technology adoption. However, the undesired side effect of free-riding is that innovators in industrialized countries lose incentives to create improvements that are most useful in developing countries, but of limited application in industrialized markets. Section 4 introduces complementarity in innovation. While innovation has no effect on the profitability of existing intermediate firms in the benchmark model, in reality new technologies can substitute or complement existing technologies. Innovation may cause technological obsolescence of previous technologies, as emphasized by Schumpeterian models. In other cases, new technologies complement rather than substitute the old ones. For instance, the market for a particular technology tends to be small at the time of its introduction, but grows as new compatible applications are developed. This complementarity in innovation can lead to multiple equilibria and poverty traps. Complementarities in the growth process may also arise from financial markets, as suggested in Section 5. The progressive endogenous enrichment of asset markets, associated with the development of new intermediate industries, may improve the diversification opportunities available to investors. This, in turn, makes savers more prepared to invest in high-productivity risky industries, thereby fostering further industrial and financial development. As a result, countries at early stages of development go through periods of slow and highly volatile growth, eventually followed by a take-off with financial deepening and steady growth. Finally, Section 6 shows how models with technological complementarities can generate rich long-run dynamics, including endogenous fluctuations between periods of high and low growth. Cycles in innovation and growth can either be due to expectational indeterminacy, or the deterministic dynamics of two-sector models with an endogenous market structure.
116
G. Gancia and F. Zilibotti
2. Growth with expanding variety In this section, we present the benchmark model of endogenous growth with expanding variety, and some extensions that will be developed in the following sections. 2.1. The benchmark model The benchmark model is a simplified version of Romer (1990), where, for simplicity, we abstract from investments in physical capital. The economy is populated by infinitely lived agents who derive utility from consumption and supply inelastic labor. The population is constant, and equal to L. Agents’ preferences are represented by an isoelastic utility function: ∞ C 1−θ − 1 U= (1) e−ρt t dt. 1−θ 0 The representative household sets a consumption plan to maximize utility, subject to an intertemporal budget constraint and a No-Ponzi game condition. The consumption plan satisfies a standard Euler equation: rt − ρ C˙ t = (2) · Ct . θ There is no physical capital, and savings are used to finance innovative investments. The production side of the economy consists of two sectors of activity: a competitive sector producing a homogeneous final good, and a non-competitive sector producing differentiated intermediate goods. The final-good sector employs labor and a set of intermediate goods as inputs. The technology for producing final goods is represented by the following production function: At 1−α α Yt = Ly,t (3) xj,t dj, 0
where xj is the quantity of the intermediate good j , At is the measure of intermediate goods available at t, Ly is labor and α ∈ (0, 1). This specification follows Spence (1976), Dixit and Stiglitz (1977) and Ethier (1982). It describes different inputs as imperfect substitutes, which symmetrically enter the production function, implying that no intermediate good is intrinsically better or worse than any other, irrespective of the time of introduction. The marginal product of each intermediate input is decreasing, and independent of the measure of intermediate goods, At . The intermediate good sector consists of monopolistically competitive firms, each producing a differentiated variety j . Technology is symmetric across varieties: the production of one unit of intermediate good requires one unit of final good, assumed to be the numeraire.2 In addition, each intermediate producer is subject to a sunk cost to 2 In Romer (1990), the variable input is physical capital, and the economy has two state variables, i.e., physical capital and knowledge.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
117
design a new intermediate input variety. New designs are produced instantaneously and with no uncertainty. The innovating firm can patent the design, and acquire a perpetual monopoly power over the production of the corresponding input. In the absence of intellectual property rights, free-riding would prevent any innovative activity. If firms could costlessly copy the design, competition would drive ex-post rents to zero. Then, no firms would have an incentive, ex-ante, to pay a sunk cost to design a new input. The research activity only uses labor. An important assumption is that innovation generates an intertemporal externality. In particular, the design of a (unit measure of) new intermediate good requires a labor input equal to 1/(δAt ). The assumption that labor productivity increases with the stock of knowledge, At , can be rationalized by the idea of researchers benefiting from accessing the stock of applications for patents, thereby obtaining inspiration for new designs. The law of motion of technical knowledge can be written as: A˙ t = δAt Lx,t ,
(4)
where δ is a parameter and Lx denotes the aggregate employment in research. The rate of technological change is a linear function of total employment in research.3 Finally, feasibility requires that L Lx,t + Ly,t . First, we characterize the equilibrium in the final good sector. Let w denote the wage, and pj be the price of the j ’th variety of intermediate input. The representative firm in the competitive final sector takes prices as parametric and chooses production and technology so as to maximize profit, given by: At At Y 1−α α πt = Ly,t (5) xj,t dj − wt Ly,t − pj,t xj,t dj. 0
0
The first-order conditions yield the following factor demands: 1−α pj,t = αLy,t xj,t α−1
and wy,t =
(1 − α)L−α y,t
∀j ∈ [0, At ]
At
xj,t α dj.
(6)
(7)
0
3 Jones (1995) generalizes this technology and lets γ (1−γ ) A˙ t = δAt A Lx,t L ,
where γA 1 is a positive externality through the stock of knowledge and γL is a negative externality that can be interpreted as coming from the duplication of research effort. Assuming γA < 1 leads to qualitative differences in the prediction of the model. In particular, the specification where γA = 1 and γL = 0, which is the model discussed here, generates scale effects. See further discussion later in this chapter and, especially, in Chapter 16 of this Handbook.
118
G. Gancia and F. Zilibotti
Next, consider the problem of intermediate producers. A firm owning a patent sets its production level so as to maximize the profit, subject to the demand function (6). The profit of the firm producing the j th variety is πj,t = pj,t xj,t − xj,t . The optimal quantity and price set by the monopolist are xj,t = xt = α 2/(1−α) Ly,t
and pj,t = p = 1/α,
(8)
respectively. Hence, the maximum profit for an intermediate producer is 1 − α 2/(1−α) Ly,t . α α Substitution of xt into (7) yields the equilibrium wage as: πj,t = πt = (p − 1)xt =
(9)
wt = (1 − α)α 2α/(1−α) At .
(10)
Next, we guess-and-verify the existence of a balanced growth (BG) equilibrium, such that consumption, production and technical knowledge grow at the same constant rate, γ , and the two sectors employ constant proportions of the workforce.4 In BG, both the production and the profits of intermediate firms, as given by Equations (8) and (9), are constant over time and across industries. Thus, xt = x and πt = π. Free entry implies that the present discounted value (PDV) of profits from innovation cannot exceed the entry cost. By the Euler equation, (2), the interest rate is also constant in BG. Hence, the PDV of profits equals π/r. The entry cost is given by the wage paid to researchers, i.e., wt /(δAt ). Therefore, the free entry condition can be written as: wt π (11) . r δAt We can then use (9) and (10), and substitute the expressions of π and wt into (11): 2/(1−α) L ( 1−α (1 − α)α 2α/(1−α) y α )α . (12) r δ The right-hand side expression is the marginal cost of innovation, independent of At , due to the cancellation of two opposite effects. On the one hand, labor productivity and, hence, the equilibrium wage grow linearly with At . On the other hand, the productivity of researchers increases with At , due to the intertemporal knowledge spillover. Thus, the unit cost of innovation is constant over time. Note that, without the externality, the cost of innovation would grow over time, and technical progress and growth would come to a halt, like in the neoclassical model. For innovation to be positive, (12) must hold with equality. We can use (i) the resource constraint, implying that Ly = L − Lx , and (ii) the fact that, from (4) and BG, Lx = γ /δ, to express (12) as a relationship between the interest rate and the growth rate:
r = α(δL − γ ).
(13)
4 The equilibrium that we characterized can be proved to be unique. Moreover, the version of Romer’s model described here features no transitional dynamics, as in AK models [Rebelo (1991)].
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
119
Figure 1.
Equation (13) describes the equilibrium condition on the production side of the economy: the higher is the interest rate that firms must pay to finance innovation expenditure, the lower is employment in research and growth. Finally, the consumption Euler equation, (2), given BG, yields: r = ρ + θγ,
(14)
which is the usual positive relation between interest rate and growth. Figure 1 plots the linear equations (13) and (14), which characterize the equilibrium. The two equations correspond, respectively, to the DD (demand for funds) and SS (supply of savings) linear schedules. An interior solution exists if and only if αδL > ρ. When this condition fails to be satisfied, all workers are employed in the production of consumption goods. When it is positive, the equilibrium growth rate is δαL − ρ , (15) α+θ showing that the growth rate is increasing in the productivity of the research sector (δ), the size of the labor force (L) and the intertemporal elasticity of substitution of consumption (1/θ ), while it is decreasing in the elasticity of final output to labor, (1 − α), and the discount rate. The trade-off between final production (consumption), on the one hand, and innovation and growth, on the other hand, can be shown by substituting the equilibrium γ =
120
G. Gancia and F. Zilibotti
expression of x into the aggregate production function, (3). This yields: Yt = α 2α/(1−α) Ly At = α 2α/(1−α) (L − γ /δ)At .
(16)
The decentralized equilibrium is inefficient for two reasons:5 1. Intermediate firms exert monopoly power, and charge a price in excess of the marginal cost of production. This leads to an underproduction of each variety of intermediate goods. 2. the accumulation of ideas produces externalities not internalized in the laissezfaire economy. Innovating firms compare the private cost of innovation, wt /(δAt ), with the present discounted value of profits, π/r. However, they ignore the spillover on the future productivity of innovation. Contrary to Schumpeterian models, innovation does not cause “creative destruction”, i.e., no rent is reduced by the entry of new firms. As a result, growth is always suboptimally low in the laissez-faire equilibrium. Policies aimed at increasing research activities (e.g., through subsidies to R&D or intermediate production) are both growthand welfare-enhancing. This result is not robust, however. Benassy (1998) shows that in a model where the return to specialization is allowed to vary and does not depend on firms’ market power (α), research and growth in the laissez-faire equilibrium may be suboptimally too high. 2.2. Two variations of the benchmark model: “lab-equipment” and “labor-for intermediates” We now consider two alternative specifications of the model that have been used in the literature, and that will be discussed in the following sections. The first specification is the so-called “lab-equipment” model, where the research activity uses final output instead of labor as a productive input.6 More formally, Equation (4) is replaced by the condition A˙ t = Yx /µ, where Yx denotes the units of final output devoted to research (hence, consumption is C = Y − Ax − Yx ) and µ the output cost per unit of innovation. In the lab-equipment model, there is no research spillover of the type discussed in the benchmark model. Labor is entirely allocated to final production (Ly = L), and the free-entry condition (12) is replaced by 1−α 2/(1−α) L α α
r
µ.
(17)
5 There is an additional reason why, in general, models with a Dixit–Stiglitz technology can generate in-
efficient allocations in laissez-faire, namely that the range of intermediate goods produced is endogenous. The standard assumption of complete markets is violated in Dixit–Stiglitz models, because there is no market price for the goods not produced. This issue is discussed in Matsuyama (1995, 1997). A dynamic example of such a failure is provided by the model of Acemoglu and Zilibotti (1997), which is discussed in detail in Section 6. 6 The “lab-equipment” model was first introduced by Rivera-Batiz and Romer (1991a); see also Barro and Sala-i-Martin (1995).
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
121
Hence, using the Euler condition, (14), we obtain the following equilibrium growth rate: γ =
(1 − α)α (1+α)/(1−α) L/µ − ρ . θ
Sustained growth is attained by allocating a constant share of production to finance the research activity. The second specification assumes that labor is not used in final production, but is used (instead of final output) as the unique input in the intermediate goods production.7 More formally, the final production technology is Yt = Z
1−α 0
At
α xj,t dj,
(18)
where Z is a fixed factor (e.g., land) that is typically normalized to unity and ignored. In this model, 1/At units of labor are required to produce one unit of any intermediate input. Therefore, in this version of the model, innovation generates a spillover on the productivity of both research and intermediate production.8 We refer to this version as the “labor-for-intermediates” model. It immediately follows that, in equilibrium, the production of each intermediate firm equals x = L−Lx . The price of intermediates is once more a mark-up over the marginal cost, pt = wt /(αAt ). In a BG equilibrium, wages and technology grow at the same rate, hence their ratio is constant. Let ω ≡ (wt /At ). The maximum profit is, then: 1−α 1−α ωx = ω(L − Lx ). π= α α The free entry condition can be expressed as: 1−α 1 (L − Lx ) , rα δ hence, γ =
1−α Lδ − ρ α
1−α +θ . α
Clearly, both the “lab-equipment” and “labor-for-intermediates” model yield solutions qualitatively similar to that of the benchmark model.
7 We follow the specification used by Young (1993). A related approach, treating the variety of inputs as consumption goods produced with labor, is examined in Grossman and Helpman (1991a). 8 The spillover on the productivity of intermediate production is not necessary to have endogenous growth. Without it, an equilibrium can be found in which production of each intermediate falls as A grows: γA = −γx . In this case, employment in production, Ax, is constant and the growth rate of Y is (1 − α)γA .
122
G. Gancia and F. Zilibotti
2.3. Limited patent protection In this section, we discuss the effects of limited patent protection. For simplicity, we focus on the lab-equipment version discussed in the previous section. The expectation of monopoly profits provides the basic incentive motivating investment in innovation; at the same time, monopoly rights introduce a distortion in the economy that raises prices above marginal costs and causes the underprovision of goods. Since the growth rate of knowledge in the typical decentralized equilibrium is below the social optimum, the presence of monopoly power poses a trade-off between dynamic and static efficiency, leading to the question, first studied by Nordhaus (1969a, 1969b), of whether there exists an optimal level of protection of monopoly rights. In the basic model, we assumed the monopoly power of innovators to last forever. Now, we study how the main results change when agents cannot be perfectly excluded from using advances discovered by others. A tractable way of doing this is to assume monopoly power to be eroded at a constant rate, so that in every instant, a fraction m of the monopolized goods becomes competitive.9 Then, for a given range of varieties in the economy, At , the number of “imitated” intermediates that have become competitive, A∗t , follows the law of motion: A˙ ∗t = m At − A∗t . (19) Stronger patent protection can be considered as a reduction in the imitation rate m. Note that the model now has two state variables, At and A∗t , and will exhibit transitional dynamics. In general, from any starting point, the ratio A∗t /At will converge to the steady-state level:10 A∗ m = , A γ +m
(20)
˙ where γ ≡ A/A. Once a product is imitated, the monopoly power of the original producer is lost and its prices is driven down to the marginal cost by competition. Thus, at each point in time, intermediates still produced by monopolists are sold as before at the markup price 1/α, while for the others, the competitive price is one. Substituting prices into demand functions yields the quantity of each intermediate sold in equilibrium: xj = α 1/(1−α) L ≡ x ∗ for j ∈ 0, A∗t , (21) xj = α 2/(1−α) L ≡ x for j ∈ A∗t , At . Note that x ∗ > x, because the monopolized goods have a higher price.
9 A growth model with limited patent life is developed by Judd (1985). Here, we follow Barro and Sala-i-
Martin (1995). An alternative way of introducing limited patent protection is to assume monopolies to have a deterministic lifetime T . In this case, the PDV of an innovation is (1−e−rT )π/r (assuming balanced growth). 10 This can be seen imposing A˙ ∗ /A∗ = γ in (19).
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
123
Free entry requires the PDV of profits generated by an innovation, V , to equal its cost µ. Along the balanced growth path, where the interest rate is constant, arbitrage in asset markets requires the instantaneous return to innovation, π/µ, to equal the real interest rate adjusted for imitation risk: r + m.11 Since prices and quantities of the monopolized goods are identical to those in the basic model, π is not affected by imitation. Imitation only affects the duration of the profit flow, which is reflected in the effective interest rate. Therefore, limiting patent lives introduces a new inefficiency: although the benefit from a discovery is permanent for the economy, the reward for the innovator is now only temporary. Using the Euler equation for consumption growth, γ = (r − ρ)/θ , and the adjusted interest rate in (17), we get the growth rate of the economy: L 1 (1 − α)α (1+α)/(1−α) − m − ρ . γ = θ µ As expected, the growth rate is decreasing in the imitation rate, as the limited duration of the monopoly effectively reduces the private value of an innovation. If we were concerned about long-run growth only, it would then be clear that patents should always be fully and eternally protected. However, for a given level of technology, At , output is higher the shorter is the patent duration (higher m), as can be seen by substituting equilibrium quantities (21) and the ratio of imitated goods (20) in the production function (3): −α/(1−α) m α Yt = α 2α/(1−α) At L 1 + −1 . γ +m Therefore, a reduction in the patent life entails a trade-off between an immediate consumption gain and future losses in terms of lower growth, and its quantitative analysis requires the calculation of welfare along the transition. Kwan and Lai (2003) perform such an analysis, both numerically and by linearizing the BG equilibrium in the neighborhood of the steady-state, and show the existence of an optimum patent life. They also provide a simple calibration, using US data on long-run growth, markups and plausible values for ρ and θ , to suggest that over-protection of patents is unlikely to happen, whereas the welfare cost of under-protection can be substantial. Alternatively, the optimal patent length can be analytically derived in models with a simpler structure. For example, Grossman and Lai (2004) construct a modified version of the model described above, where they assume quasi-linear functions. They show the 11 A simple way of seeing this is through the following argument. In a time interval dt, the firm provides a
profit stream π · dt, a capital gain of V˙ · dt if not imitated and a capital loss V if imitated (as the value of the patent would drop to zero). In the limit dt → 0, the probability of being imitated in this time interval is m · dt and the probability of not being imitated equals (1 − m · dt). Therefore, the expected return for the firm is π · dt + (1 − m · dt)V˙ · dt − mV · dt. Selling the firm and investing the proceeds in the capital market would yield an interest payment of rV · dt. Arbitrage implies that the returns from these two forms of investment should be equal and in a steady state V˙ = 0, implying π/V = r + m.
124
G. Gancia and F. Zilibotti
optimal patent length to be an increasing function of the useful life of a product, of consumers’ patience and the ratio of consumers’ and producers’ surplus under monopoly to consumers’ surplus under competition. In addition, they derive the optimal patent length for noncooperative trading countries and find that advanced economies with a higher innovative potential will, in general, grant longer patents. A similar point is made in Lai and Qiu (2003).
3. Trade, growth and imitation Growth models with an expanding variety of products are a natural dynamic counterpart to the widely-used trade models based on increasing returns and product differentiation developed in the 1980s [e.g., Helpman and Krugman (1985)]. As such, they offer a simple framework for studying the effects of market integration on growth and other issues in dynamic trade theory. Quality-ladder models have also been proposed in this literature, but they are a less natural counterpart to the static new trade theory, as they do not focus on the number of varieties available in an economy and their growth rate. As we shall see, economic integration can provide both static gains, through the access to a wider range of goods, and dynamic gains, through an increase in the rate at which new varieties are introduced. However, the results may vary when integration is limited to commodity markets with no international diffusion of knowledge [Rivera-Batiz and Romer (1991a)] and when countries differ in their initial stock of knowledge [Devereux and Lapham (1994)]. Finally, the analysis in this section is extended to product-cycle trade: the introduction of new products in advanced countries and their subsequent imitation by less developed countries. An important result will be to show that, contrary to the closed economy case, imitation by less developed countries may spur innovation and growth [Helpman (1993)]. 3.1. Scale effects, economic integration and trade In this section, we use the benchmark model to discuss the effects of trade and integration. The model features scale effects. Take two identical countries with identical labor endowment, L = L∗ . In isolation, both countries would grow at the same rate, as given by (15). But if they merge, the growth rate of the integrated country increases to: γI =
δα(L + L∗ ) − ρ 2αδL − ρ = . α+θ α+θ
Therefore, the model predicts that economic integration boosts growth. Integration, even if beneficial, may be difficult to achieve. However, in many instances, trade operates as a substitute for economic integration. Rivera-Batiz and Romer
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
125
(1991a) analyze under which condition trade would attain the same benefits as economic integration. To this aim, they consider two experiments:12 1. The economies can trade at no cost in goods and assets, but knowledge spillovers remain localized within national borders; 2. In addition, knowledge spillovers work across borders after trade. In both cases, to simplify the analysis, the two economies are assumed to produce, before trade, disjoint subsets of intermediate goods. This assumptions avoids complications arising from trade turning monopolies into duopolies in those industries which exist in both countries. Clearly, after trade, there would be no incentive for overlap in innovation, and the importance of inputs that were historically produced in both countries would decline to zero over time. We start from the case analyzed by Rivera-Batiz and Romer (1991a), where the two countries are perfectly identical before trade. Namely, L = L∗ and A0 = A∗0 , where the star denotes the foreign economy, and time zero denotes the moment when trade starts. Since, in a BG equilibrium, γ = δLx , trade can only affect growth via the split of the workforce between production and research. Such a split, however, is not affected by trade, for in the symmetric equilibrium, trade increases the productivity of workers in production and the profitability of research by the same proportion. Since both the cost and private benefit of innovation increase by the same factor, investments in innovation remain unchanged. More formally, the after trade wage is α ∗ wtrade = (1 − α)L−α (22) y x A+A , which is twice as large as in the pre-trade equilibrium since at the moment of trade liberalization, A = A∗ . Higher labor costs are a disincentive to research. But trade also increases the market for intermediate goods. Each monopolist can now sell its product in two markets. Since the demand elasticity is the same in both markets, the monopoly price equals 1/α in both markets. Thus, the after trade profit is 1 − α 2/(1−α) α Ly . α The free-entry condition becomes, for both countries: πtrade = (p − 1)(x + x ∗ ) = 2
1−α 2/(1−α) Ly α α
(1 − α)α 2α/(1−α) (23) , r δ which, after simplifying, is identical to (12). Therefore, the split of the workforce between production and research remains unchanged, and trade has no permanent effects on growth. Opening up to free trade, however, induces a once-and-for-all gain: both output and consumption increase in both countries, similarly to an unexpected increase 2
2
12 The original article considers two versions of the model, one using the benchmark set-up and the other
using the “lab-equipment” version. For the sake of brevity, we restrict the attention to the first. Romer (1994) extends the analysis to the case when a tariff on imports is imposed.
126
G. Gancia and F. Zilibotti
in the stock of knowledge, since final producers in both countries can use a larger set of intermediate goods. This result is not robust to asymmetric initial conditions. Devereux and Lapham (1994) show that if, initially, the two countries have different productivity levels, trade leads to specialization and a rise in the world growth rate.13 Consider the economies described above, but assume that A0 < A∗0 . Recall that free-entry implies: w w∗ and V ∗ , δA δA∗ where V , V ∗ denote the PDV of profits for an intermediate firm located at home and abroad, respectively. First, trade in intermediate goods and free capital markets equalize the rate of return to both financial assets (r) and labor (w).14 Second, monopoly profits are independent of firms’ locations, thereby implying that the value of firms must be the same all over the world: V = V ∗ = V w . Therefore, at the time of trade liberalization, we must have: wtrade wtrade V w, > δA δA∗ implying that no innovation is carried out in equilibrium in the (home) country, starting from a lower productivity. Moreover, the productivity gap in R&D widens over time: indeed, trade forever eliminates the incentives to innovate in the initially poorer country. In the richer (foreign) country, however, trade boosts innovation.15 The value of foreign firms must satisfy the following Bellman equation: V
1 − α 2/(1−α) ∗ α Ly + L , α where we note that Ly = L. The free-entry condition implies that: rV ∗ = V˙ ∗ +
(1 − α)α 2α/(1−α) A∗ + A . δ A∗ Since knowledge only accumulates in the foreign country, the value of intermediate firms must decline over time, and in the long-run tend to its pre-trade value, i.e., V ∗ = (1 − α)α 2α/(1−α) /δ. Therefore, in the long run, the free-entry condition is V∗ =
1−α 2/(1−α) (L + L∗y ) α α
r
(1 − α)α 2α/(1−α) . δ
(24)
13 See also Rivera-Batiz and Romer (1991b) on the effects of trade restrictions with asymmetric countries. 14 Recall Equation (22). The equalization of wages descends from a particular feature of the equilibrium, i.e.,
that the marginal product of labor is independent of the level of employment in production (since x is linear in Ly ). This feature is not robust. If the production technology had land as an input, for instance, wages would not be equalized across countries; see Devereux and Lapham (1994) for an analysis of the more general case. 15 Our discussion focuses on a world where no economy becomes fully specialized in research, since this seems to be the empirically plausible case.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
127
Comparing (24) with (12) shows that trade reduces employment in production and, consequently, increases the long-run research activity in the foreign country, which implies that trade increases growth. In terms of Fig. 1, trade creates an outward shift in the DD schedule, leading to a higher interest rate and faster growth in equilibrium. The result can be interpreted as trade leading to specialization. The home country specializes in final production, while the foreign country diversifies between manufacturing and innovation.16 This is efficient, since there are country-wide economies of scale in innovation. Although trade leads to zero innovation in the home country, markets are integrated: final good producers, in both countries, can use the same varieties of intermediates and all consumers in the world can invest in the innovative firms of the foreign economy. Therefore, the location of innovation and firms has no impact on the relative welfare of the two countries. Consider now the case when trade induces cross-country flows of ideas, i.e., if the knowledge spillover is determined, after trade, by the world stock of ideas contained in the union of A and A∗ . When free trade is allowed, the accumulation of knowledge in each country is given by A˙ = δLx A + A∗ and A˙ ∗ = δL∗x A + A∗ . Even if trade did not affect the allocation of the workforce between production and research, the rate of growth of technology would increase. But there is an additional effect; the larger knowledge spillover increases labor productivity in research, inducing an increase of employment in research. Formally, the total effect is equivalent to an increase in parameter δ. In terms of Figure 1, trade in goods plus flow of ideas imply an upward shift of the DD locus for both countries. Hence, trade attains the same effect as economic integration (increasing δ is equivalent to increasing L). This result is robust to asymmetric initial conditions. 3.2. Innovation, imitation and product cycles The model just presented may be appropriate for describing trade integration between similar countries, but it misses important features of North–South trade. In a seminal article, Vernon (1966) argued that new products are first introduced in rich countries (the North), where R&D capabilities are high and the proximity to large and rich markets facilitates innovation. After some time, when a product reaches a stage of maturity and manufacturing methods become standardized, the good can easily be imitated and then, the bulk of production moves to less developed countries (the South), to take advantage of low wages. The expanding variety model provides a natural framework for studying the introduction of new goods and their subsequent imitation (product cycle trade).17 16 Home-country patent holders will still produce intermediates, but as compared to the world’s stock of
intermediates, they will be of measure zero. 17 Quality ladder models of innovation have been used to study product cycles by, among others, Grossman
and Helpman (1991a), Segerstrom et al. (1990) and Dinopoulos and Segerstrom (2003).
128
G. Gancia and F. Zilibotti
We have already discussed imitation within the context of a closed economy. Here, we extend the analysis to the case where a richer North innovates, while a poorer South only engages in imitation. The analysis yields new results that modify some of the previous conclusions on the effect of imitation on innovation. The key questions are, first, how the transfer of production to the South through imitation affects the incentives to innovate and, second, how it affects the income distribution between North and South. Following Helpman (1993), consider a two-region model of innovation, imitation and trade. Assume that R&D, producing new goods, is performed in the North only and that costless imitation takes place in the South at a constant rate m.18 The imitation rate can be interpreted as an inverse measure of protection of Intellectual Property Rights (IPRs). Once a good is copied in the South, it is produced by competitive firms. Therefore, at every point in time, there is a range AN t of goods produced by monopolists in the North and a range ASt of goods that have been copied and are produced in the South by competitive firms. Given that the rate of introduction of new good is γ = A˙ t /At , S where At = AN t + At , and that monopolized goods are copied at the instantaneous rate S N S ˙ m, At = mAt , it follows that a steady-state where the ratio AN t /At is constant must satisfy: γ AN = A γ +m
and
AS m = . A γ +m
(25)
We use the “labor-for-intermediates” version of the growth model, so that the price of a single variety depends on the prevailing wage rate in the country where it is manufactured. This is an important feature of product cycle models, allowing the North to benefit from low production costs in the South for imitated goods. Therefore, we define the aggregate production function as in (18): At Yt = (26) xi α di, 0
where At is the (growing) range of available products xi and ε = 1/(1−α) is the elasticity of substitution between any two varieties. Intermediates are manufactured with 1/At units of labor per unit of output in both regions. Northern firms charge a monopoly price, as long as their products have not been imitated, equal to a constant markup 1/α over the production cost, given by the wage rate. On the contrary, Southern firms produce imitated goods that have become competitive and sell them at a price equal to the marginal cost. To summarize: ptN =
wtN αAt
and ptS =
wtS , At
(27)
where ptN and ptS are the prices of any variety of intermediates produced in the North and South, respectively. 18 The rate of imitation is made endogenous in Grossman and Helpman (1991b).
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
129
As in the benchmark model, innovation requires labor: the introduction of new products per unit of time A˙ t equals δAt Lx , where Lx is the (Northern) labor input employed in R&D, δ is a productivity parameter and At captures an externality from past innovations. This implies that the growth rate of the economy is a linear function of the number of workers employed in R&D, γ = Lx δ. As usual, profits generated by the monopoly over the sale of the new good are used to cover the cost of innovation. Since the profits per product are a fraction (1 − α) of total revenue p N x and the labor market clears, N AN t x/At + γ /δ = L , profits can be written as: 1 − α wtN γ N N π = (28) L − . α AN δ t Arbitrage in asset markets implies that (r + m)V N = π N + V˙ N , where V N is the PDV of a new good and the effective interest rate is adjusted by the imitation risk. Along a BG path, V˙ N = 0 and free entry ensures that the value of an innovation equals its cost, wtN /δAt . Combining these considerations with (25) and (28) yields: γ + m 1 − α N δL − γ = r + m. α γ
(29)
Together with the Euler equation for consumption growth, (29) provides an implicit solution for the long-run growth rate of innovation. Note that the left-hand side is the profit rate (i.e., instantaneous profits over the value of the innovation) and the right-hand side represents the effective cost of capital, inclusive of the imitation risk. To see the effect of a tightening of IPRs (a reduction of m), consider how an infinitesimal change in m affects the two sides of (29). Taking a log linear approximation, the impact of m on the profit rate is 1/(γ + m), whereas the effect on the cost of capital is 1/(r + m). In the case of log preferences, studied by Helpman (1993), r > γ . Hence, a reduction of m has a larger impact on the profit rate than on the effective cost of capital, thereby reducing the profitability of innovation and growth. What is the effect on the fraction of goods produced in the North? Rewriting (29) with the help of (25) as: 1 1 − α N AN = δL − γ , (30) A α r +m it becomes apparent that a reduction of m increases the share of goods manufactured in the North, both through its direct effect and by reducing γ and r. To understand these results, note that stronger IPRs have two opposite effects. First, a lower imitation rate prolongs the expected duration of the monopoly on a new product developed in the North, thereby increasing the returns to innovation. Second, since firms produce for a longer time in the North, it rises the demand for Northern labor, w N , and hence, the cost of innovation. For the specification with log utility, the latter effect dominates and innovation declines. More generally, the link between the rate of imitation and innovation can go either way [as in Grossman and Helpman (1991a)]. However, the important result here is that tighter IPRs does not necessarily stimulate innovation in the long run.
130
G. Gancia and F. Zilibotti
The effect of IPRs on the North–South wage ratio can be found using (27), together with the relative demand for intermediates:19 N 1/ε A LS wN =α . (31) wS AS LN − γ /δ Given that a decline in the imitation rate m raises (AN /AS )/(LN − γ /δ) (see Equation (30)), a tightening of IPRs raises the relative wage of the North. Helpman (1993) computes welfare changes in the North and in the South (including transitional dynamics) after a change in the imitation rate m, and concludes that the South is unambiguously hurt by a decline in imitation. Moreover, if the imitation rate is not too high, the North can also be worse-off. More recent papers on product cycles, incorporating the notion that stronger IPRs make relocation of production to the South a more attractive option, have come to different conclusions. For example, by assuming that Northern multinationals can produce in the South and that Southern firms can only imitate after production has been transferred to their country, Lai (1998) shows that stronger IPRs increase the rate of product innovation and the relative wage of the South. Similarly, Yang and Maskus (2001) find that if Northern firms can license their technology to Southern producers, being subject to an imitation risk, stronger IPRs reduce the cost of licensing, free resources for R&D and foster growth, with ambiguous effects on relative wages. Finally, the literature on appropriate technology [e.g., Diwan and Rodrik (1991), Acemoglu and Zilibotti (2001), Gancia (2003)] has shown that, when the North and the South have different technological needs, the South has an incentive to protect IPRs in order to attract innovations more suited to their technological needs. Some of these results are discussed in the next sections.
4. Directed technical change So far, technical progress has been modeled as an increase in total factor productivity (A) that is neutral towards different factors and sectors. For many applications, however, this assumption is not realistic. For example, there is evidence that technical progress has been skill-biased during the last century and that this bias accelerated during the 1980s. Similarly, the fact that the output shares of labor and capital have been roughly constant in the US while the capital–labor ratio has been steadily increasing
19 Relative demand for intermediates is:
pN = pS
N α−1 x . xS
N S S S Using x N = ALN y /A , x = ALy /A and the pricing formula (27) yields the expression in the text.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
131
suggests that technical change has mainly been labor-augmenting.20 Further, industry studies show R&D intensity to vary substantially across sectors. In order to build a theory for the direction of technical change, a first step is to introduce more sectors into the model. Then, studying the economic incentives to develop technologies complementing a specific factor or sector can help understand what determines the shape of technology. An important contribution of this new theory will be to shed light on the determinants of wage inequality [Acemoglu (1998, 2003a)]. Another application studies under which circumstances technologies developed by profit-motivated firms are appropriate for the economic conditions of the countries where they are used. The analysis will demonstrate that, since IPRs are weakly protected in developing countries, new technologies tend to be designed for the markets and needs of advanced countries. As a result, these technologies yield a low level of productivity when adopted by developing countries [Acemoglu and Zilibotti (2001)]. Trade can reinforce this problem and create interesting general equilibrium effects. Although most of the results discussed in this section can be derived using models of vertical innovation, the expanding variety approach has proved to be particularly suited for addressing these issues because of its analytical tractability and simple dynamics. For instance, creative destruction, a fundamental feature of quality-ladder models, is not a crucial element for the problems at hand, and abstracting from it substantially simplifies the analysis. 4.1. Factor-biased innovation and wage inequality Directed technical change was formalized by Acemoglu (1998), and then integrated by Acemoglu and Zilibotti (2001) into a model of growth with expanding variety, to explain the degree of skill-complementarity of technology.21 In this section, we discuss the expanding variety version [following the synthesis of Acemoglu (2002)] by extending the “lab-equipment” model to two sectors employing skilled and unskilled labor, respectively. Consider the following aggregate production function: (ε−1)/ε (ε−1)/ε ε/(ε−1) + YH , Y = YL (32) where YL and YH are goods produced with unskilled labor, L, and skilled labor, H , respectively. Y represents aggregate output, used for both consumption and investment, as a combination of the two goods produced in the economy, with an elasticity of substitution equal to ε. Maximizing Y under a resource constraint gives constant elasticity demand functions, implying a negative relationship between relative prices and relative
20 Unless the production function is Cobb–Douglas, in which case the direction of technical progress is
irrelevant. Empirical estimates suggest that the elasticity of substitution between labor and capital is likely to be less than one. See Hamermesh (1993) for a survey of early estimates and Krusell et al. (2000) and Antras (2004) for more recent contributions. 21 Important antecedents are Kennedy (1964), Samuelson (1965) and Atkinson and Stiglitz (1969).
132
G. Gancia and F. Zilibotti
quantities: PH YL 1/ε = , PL YH
(33)
where PL and PH are the prices of YL and YH , respectively. Aggregate output is chosen as the numeraire, hence: 1/(1−ε) 1−ε = 1. PL + PH1−ε (34) The distinctive feature of this model is that the two goods are now produced using different technologies: AL YL = L1−α xL,j α dj, 0 (35) AH xH,j α dj, YH = H 1−α 0
where xL,j , j ∈ [0, AL ], are intermediate goods complementing unskilled labor, whereas xH,j , j ∈ [0, AH ] complement skilled labor. This assumption captures the fact that different factors usually operate with different technologies and that a new technology may benefit one factor more than others.22 For example, it has been argued that computers boosted the productivity of skilled more than that of unskilled labor, whereas the opposite occurred after the introduction of the assembly line. As before, technical progress takes the form of an increase in the number of intermediate goods, [AL , AH ], but now an innovator must decide which technology to expand. The profitability of the two sectors pins down, endogenously, the direction of technical change. In a steadystate equilibrium, there is a constant ratio of the number of intermediates used by each factor, AH /AL , and this can be interpreted as the extent of the “endogenous skill-bias” of the technology. The analysis follows the same steps as in model with a single factor. Final good producers take the price of their output (PL , PH ), the price of intermediates (pL,j , pH,j ) and wages (wL , wH ) as given. Consider a variety j used in the production of YL . Profit maximization gives the following isoelastic demand: αPL 1/(1−α) xL,j = (36) L, pL,j and an equivalent expression for xH,j . The intermediate good sector is monopolistic, with each producer owning the patent for a single variety. The cost of producing one unit of any intermediate good is one
22 The analysis can be generalized to specifications where, in the spirit of Heckscher–Ohlin models, each
sector uses all productive factors, but factor intensities differ across sectors. The model can also be generalized to more than two factors and sectors.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
133
unit of the numeraire. The symmetric structure of demand and technology implies that all monopolists set the same price, pLj = pL . In particular, given the isoelastic demand, they set pL = 1/α and sell the quantity xL,j = (α 2 PL )1/(1−α) L. The profit flow accruing to intermediate producers can therefore be expressed as πL = (1 − α)α (1+α)/(1−α) (PL )1/(1−α) L.
(37)
Similar conclusions are reached for varieties used in the production of YH , leading to πH = (1 − α)α (1+α)/(1−α) (PH )1/(1−α) H.
(38)
From (37)–(38), it immediately follows that the relative profitability in the two sectors is given by PH 1/(1−α) H πH , = (39) πL PL L which, since profits are used to finance innovation, is also the relative profitability of R&D directed to the two sectors. The first term in (39) represents the “price effect”: there is a greater incentive to invent technologies producing more expensive goods.23 The second term is the “market size” effect: the incentive to develop a new technology is proportional to the number of workers that will be using it.24 Next, using the price of intermediates in (36) and (35) gives final output in each sector: α/(1−α)
YL = α 2α/(1−α) PL YH = α
2α/(1−α)
AL L, α/(1−α) PH AH H.
(40)
Note the similarity with (16). As in the benchmark model, output – in each sector – is a linear function of technology and labor. But sectoral output now also depends on sectoral prices, PL and PH , since a higher price of output increases the value of productivity of intermediates, but not their costs, and therefore encourages firms to use more of them, thereby raising labor productivity. Note that this is not the case in the one-sector model since there, the price of output is proportional to the price of intermediates. We can now solve for prices and wages as functions of the state of technology and endowments. Using (40) into (33) and noting that the wage bill is a constant fraction of sectoral output, yields: AH H −(1−α)/σ PH = , (41) PL AL L wH AH 1−1/σ H −1/σ (42) = , wL AL L 23 The price effect, restated in terms of factor prices, was emphasized by Hicks (1932) and Habakkuk (1962). 24 Market size, although in the context of industry- and firm-level innovation, was emphasized as a determi-
nant of technical progress by Griliches and Schmookler (1963), Schmookler (1966) and Schumpeter (1942).
134
G. Gancia and F. Zilibotti
where σ ≡ 1 + (1 − α)(ε − 1) is, by definition, the elasticity of substitution between H and L.25 Note that the skill premium, wH /wL , is decreasing in the relative supply of skilled labor (H /L) and increasing in the skill-bias (AH /AL ) as long as σ > 1. The final step is to find the equilibrium for technology. We assume, as in the labequipment model of Section 2.2, the development of a new intermediate good to require a fixed cost of µ units of the numeraire. Free entry and an arbitrage condition require the value Vz of an innovation directed to factor Z ∈ {L,H } to equal its cost. Since the value of an innovation is the PDV of the infinite stream of profits it generates, an equilibrium with a positive rate of innovation in both types of intermediates such that the ratio AH /AL remains constant, i.e., a BG path where PH /PL , wH /wL and πH /πL are also constant, requires profit equalization in the two sectors, πH = πL = π. Imposing this restriction yields the equilibrium skill-bias of technology, σ −1 AH H (43) = . AL L Equation (43) shows that, as long as workers of different skill levels are gross substitutes (σ > 1), an increase in the supply of one factor will induce more innovation directed to that specific factor. This is the case because, with σ > 1, the market size effect dominates the price effect, and technology is biased towards the abundant factor. The opposite is true if σ < 1. As usual, the growth rate of the economy can be found from the free-entry condition πZ /r = µ, Z ∈ {L,H }. Using (34), (41) and (43) to substitute for prices and the interest rate from the Euler equation, yields: 1 (1 − α)α (1+α)/(1−α) σ −1 σ −1 1/(σ −1) L +H −ρ . γ = θ µ If we only had one factor (e.g., H = 0), the growth rate would reduce to that of the benchmark model. Directed technical change has interesting implications on factor prices. Using (43), the skill-premium becomes: σ −2 wH H (44) = . wL L Equation (44) shows that the slope of the labor demand curve, i.e., the relationship between relative wages and relative labor supply, can be either positive or negative and is the result of two opposite forces. On the one hand, a large supply of one factor depresses the price of its product while, on the other hand, it induces a technology bias in its favor, thereby raising its productivity. A high substitutability between H and L implies a weak price effect of an increase in relative supply, which makes a positive relationship more likely. In particular, if σ > 2, the market size effect is sufficiently strong to not
25 This is the short-run elasticity of substitution between L and H , for a given technology A and A . L H
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
135
only dominate the price effect on technical change (see Equation (43)), but also the substitution effect between skilled and unskilled workers at a given technology. This result can help rationalize several facts. First, it suggests that technical change has been skill biased during the past 60 years, because of the steady growth in the supply of skilled labor. Second, the case σ > 2 offers an explanation for the fall and rise in the US skill premium during the 1970s and 1980s. In the 1970s, there was a large increase in the supply of skilled labor (H /L). Assuming this shock to be unexpected, the model predicts an initial fall in the skill premium (recall that AH /AL is a state variable that does not immediately adjust), followed by its rise due to the induced skill biased technical change, a pattern broadly consistent with the evidence. In Acemoglu (2003b), this set-up is used to study the direction of technical progress when the two factors of production are capital and labor. Beyond the change of notation, the resulting model has an important qualitative difference, as capital can be accumulated. The main finding is that, when both capital and labor augmenting innovations are allowed, a balanced growth path still exists and features labor-augmenting technical progress only. The intuition is that, while there are two ways of increasing the production of capital-intensive goods (capital-augmenting technical change and accumulation), there is only one way of increasing the production of labor-intensive goods (laboraugmenting technical progress). Therefore, in the presence of capital accumulation, technical progress must be more labor-augmenting than capital-augmenting. Further, Acemoglu shows that, if capital and labor are gross complements (i.e., the elasticity of substitution between the two is less than one), which seems to be the empirically relevant case [see, for example, Antras (2004)], the economy converges to the balanced growth path. Finally, the theory of directed technical change can be used to study which industries attract more innovation and why R&D intensity differs across sectors. In this exercise, following a modified version of Klenow (1996), we abstract from factor endowments as determinants of technology, by assuming there to be a single primary input, which we call labor. Instead, other characteristics can make one sector more profitable than others. Major explanations put forward in the literature on innovation are industry differences in technological opportunities, market size and appropriability of rents, all factors that can easily be embedded in the basic model with two sectors. In particular, to capture the market size hypothesis, we introduce a parameter η defining the relative importance of industry i in aggregate consumption: (ε−1)/ε (ε−1)/ε ε/(ε−1) Y = ηYi + (1 − η)Yj . Differences in technological opportunities can be incorporated by allowing the cost of an innovation, µi , to vary across sectors. Finally, we assume that an inventor in industry i can only extract a fraction λi of the profits generated by his innovation. The previous analysis carries over almost unchanged, with the main difference that we now need to solve for the allocation of labor across industries. This can be done requiring all
136
G. Gancia and F. Zilibotti
industries to pay the same wage, i.e., setting (42) equal to one: σ −1 Ai Li = . Lj Aj Solving the new arbitrage condition stating that innovation for the two industries should be equally profitable in BG, λi πi /µi = λj πj /µj , yields the relative industry-bias of technology: ε/(2−σ ) η Ai λi µj 1/(2−σ ) = . Aj λj µi 1−η As expected, industries with a larger market size, better technological opportunities and higher appropriability attract more innovations.26 Empirical estimates surveyed by Cohen and Levin (1989) suggest that about one half of the industry differences in research intensity can be attributed to the available measures of these three factors. 4.2. Appropriate technology and development Directed technical change has interesting implications for the analysis of some development issues. Acemoglu and Zilibotti (2001) show that technologies resulting from directed technical change are optimal for the economic conditions of the markets where they are sold. They analyze the implications of this finding in a two-country world where technological innovation takes place in the North, and the South does not enforce (or imperfectly enforce) IPRs. In this environment, innovators in the North can only extract rents from selling technologies (embodied in new varieties of intermediate goods) in the Northern market, since new technologies can be copied and locally produced in the South. Thus, innovation does not respond to the factor endowment of the South: the equilibrium skill-bias of technical change (see Equation (43) in the previous section) is determined by the factor endowment of the North only. In this sense, technological development tends to be “inappropriate” for the South: there is too much investment in inventing new technologies augmenting the productivity of skilled workers, and too little in inventing new technologies augmenting the productivity of unskilled workers. Such excessive skill-bias prevents the South from fully profiting from technological improvements. The theory can explain North–South productivity differences, even when the technology is identical and there are no significant barriers to technology adoption.27 26 This is true as long as σ < 2. This restriction is required to have balanced growth. If violated, i.e., if goods
are highly substitutable, it would be profitable to direct innovation to one sector only. 27 Evidence on cross-country TFP differences is provided by, among others, Klenow and Rodriguez-Clare
(1997), Hall and Jones (1999), Caselli, Esquivel and Lefort (1996) and Prescott (1998). The view that technological differences arise from barriers to technology adoption is expressed by, among others, Parente and Prescott (1994) and Prescott (1998).
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
137
We start by studying the set of advanced countries, called North. A continuum of measure one of final goods is produced by competitive firms. Final goods, indexed by 1 i ∈ [0,1], are aggregated to give a composite output, Y = exp( 0 log yi di), which is the numeraire. There are two differences with respect to the model of the previous section: first, there is a continuum of sectors, not just two, and second, the elasticity of substitution between sectors is unity.28 Each good i can be produced with both skilled and unskilled labor using two sets of intermediate goods: intermediates [0,AL ] used by unskilled workers only and intermediates [0,AH ] used by skilled workers only. Therefore, despite the continuum of sectors, there are only two types of technologies, as in the basic model of directed technical change. The production function takes the following form: AH
1−α AL α 1−α α yi = (1 − i)li (45) xL,v,i dv + [ihi ] xH,v,i dv, 0
0
where li and hi are the quantities of unskilled and skilled labor employed in sector i, respectively, and xz,v,i is the quantity of intermediate good of type v used in sector i together with the labor of skill level z = L, H . Note that sectors differ in laboraugmenting productivity parameters, (1 − i) for the unskilled technology and i for the skilled technology, so that unskilled labor has a comparative advantage in sectors with a low index. Producers of good i take the price of their product, Pi , the price of intermediates (pL,v , pH,v ) and wages (wL , wH ) as given. Profit maximization gives the following demands for intermediates: xL,v,i = (1 − i)li [αPi /pL,v ]1/(1−α)
and xH,v,i = ihi [αPi /pH,v ]1/(1−α) . (46)
The intermediate good sector is monopolistic. Each producer holds the patent for a single type of intermediate good v, and sells its output to firms in the final good sectors. The cost of producing one unit of any intermediate is conveniently normalized to α 2 units of the numeraire. Profit maximization by monopolists implies that prices are a constant markup over marginal costs, p = α. Using the price of intermediates together with (46) and (45) gives the final output of sector i as a linear function of the number of intermediate goods and labor:
α/(1−α) yi = Pi (47) AL (1 − i)li + AH ihi . From (47), it is easily seen that all sectors whose index i is below a threshold level J will use the unskilled technology only and the remaining sectors will employ the skilled technology only. This happens because of the comparative advantage of unskilled workers in low index sectors and the linearity of the production function (there is no incentive to combine the two technologies and, for a given i, one always dominates the other).
28 The composite output Y can be interpreted as a symmetric Cobb–Douglas over the measure of final goods
i ∈ [0,1].
138
G. Gancia and F. Zilibotti
The total profits earned by monopolists are:29 J 1/(1−α) Pi (1 − i)li di πL,v = (1 − α)α
and
0
πH,v = (1 − α)α
J
1
(48) 1/(1−α) Pi ihi
di.
Note that, by symmetry, πL,v = πL,j and πH,v = πH,j . Given the Cobb–Douglas specification in (45), the wage bill in each sector is a fraction (1 − α) of sectoral output. Therefore, Equation (47) can be used to find wages:30 wL = (1 − α)Pi 1/(1−α) AL (1 − i)
and wH = (1 − α)Pi 1/(1−α) AH i.
(49)
Defining PL ≡ P0 , PH ≡ P1 and dividing equations in (49) by their counterparts in sectors 0 and 1, respectively, it is possible to derive the following pattern of prices: for i J , Pi = PL (1 − i)−(1−α) and for i J , Pi = PH i −(1−α) . Intuitively, the price of a good produced with skilled (unskilled) labor is decreasing in the sectoral productivity of skilled (unskilled) workers. Next, note that to maximize Y , expenditures across goods must be equalized, i.e., Pi yi = PH y1 = PL y0 (as for a symmetric Cobb– Douglas). This observation, plus the given pattern of prices and full employment, imply that labor is evenly distributed among sectors: li = L/J , hi = H /(1 − J ), as prices and sectoral productivity compensate each other. Finally, in sector i = J , it must be the case that both technologies are equally profitable or PL (1 − J )−(1−α) = PH J −(1−α) ; this condition, using PH y1 = PL y0 and (47), yields: PH 1/(1−α) AH H −1/2 J = = . (50) 1−J PL AL L The higher the relative endowment of skill (H /L) and the skill-bias of technology (AH /AL ), the larger the fraction of sectors using the skill-intensive technology (1 − J ). Finally, integrating Pi yi over [0, 1], using (47), (50) and the fact that the consumption 1 aggregate is the numeraire (i.e., exp[ 0 ln Pi di] = 1) gives a simple representation for aggregate output:
2 Y = exp(−1) (AL L)1/2 + (AH H )1/2 , (51) which is a CES function of technology and endowments, with an elasticity of substitution between factors equal to two.
29 Note that the integral in the right-hand side expression of π L,υ (πH,υ ) ranges from 0 to J (from J to 1),
since the producers of unskilled (skilled) intermediate goods sell their products to the J (to the 1 − J ) final industries using the unskilled (skilled) technology. 30 In Acemoglu and Zilibotti (2001), there is an additional parameter (Z > 1), which is here omitted for simplicity, which augments the productivity of skilled workers, ensuring that the skill premium is positive in equilibrium.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
139
So far, the analysis defines an equilibrium for a given technology. Next, we need to study innovation and characterize the equilibrium skill-bias of technology, (AH /AL ). As before, technical progress takes the form of an increase in AL and AH and is the result of directed R&D investment. The cost of an innovation (of any type) is equal to µ units of the numeraire, and R&D is profitable as long as the PDV of the infinite flow of profits that a producer of a new intermediate expects to earn covers the fixed cost of innovation. Finally, free entry ensures that there are no additional profits. Using the price pattern, instantaneous profits can be simplified as: 1/(1−α)
πH = α(1 − α)PH
H.
(52)
A parallel expression gives πL . Balanced growth requires πL = πH ; in this case, AH and AL grow at the same rate, the ratio AH /AL is constant as are J , PL and PH . Imposing πL = πH in (52) and using (50) yields: H 1−J AH = . = AL J L
(53)
Note that the equilibrium skill-bias is identical to that of (43) in the special case when σ = 2. Further, (53) shows that the higher is the skill endowment of a country, the larger is the range of sectors using the skilled technology. This is a complete characterization of the equilibrium for fully integrated economies developing and selling technologies in their markets with full protection of IPRs and can be interpreted as a description of the collection of rich countries, here called the North. Consider now Southern economies, where skilled labor is assumed to be relatively more scarce: H S /LS < H N /LN . Assume that intellectual property rights are not enforced in the South and that there is no North–South trade. It follows that intermediate producers located in the North cannot sell their goods or copyrights to firms located in the South, so that the relevant market for technologies is the Northern market only. Nonetheless, Southern producers can copy Northern innovations at a small but positive cost. As a consequence, no two firms in the South find it profitable to copy the same innovation and all intermediates introduced in the North are immediately copied (provided that the imitation cost is sufficiently small) and sold to Southern producers by a local monopolist. Under these assumptions, firms in the South take the technologies developed in the North as given and do not invest in innovation.31 This means that both the North and the South use the same technologies, but AH /AL = H N /LN , i.e., the skill-bias is determined by the factor endowment of the North, since this is the only market for new technologies. Except for this, the other equilibrium conditions also apply to the South after substituting the new endowments, H S and LS .
31 Imitation can be explicitly modelled as an activity similar to innovation, but less costly. Assuming the cost of an innovation of type z to decrease with the distance from the relevant technology frontier AN Z , as in Barro
and Sala-i-Martin (1997), would yield very similar results.
140
G. Gancia and F. Zilibotti
We are now ready to answer the following questions: are technologies appropriate for the skill endowment of the countries where they are developed? What happens to aggregate productivity if they are used in a different economic environment? Simple differentiation on (51) establishes that Y is maximized for AH /AL = H /L. This is exactly condition (53), showing that the equilibrium skill-bias is optimally chosen for the Northern skill composition. On the contrary, since factor abundance in the South does not affect the direction of technical change, new technologies developed in the North are inappropriate for the needs of the South. As a consequence, output per capita, Y/(L + H ) is greater in the North than in the South. The reason for these productivity differences is a technology–skill mismatch. To understand why, note that, from Equation (50), J S > J N . Rewriting (53) as AH J N = AL (1 − J N ) and inspecting Equation (47) reveals that unskilled workers are employed in the North up to sector J N , where they become as productive as skilled workers. This basic efficiency condition is violated in the South, where AH J S > AL (1 − J S ). Because of its smaller skill endowment, the South is using low-skill workers in some sectors where high-skill workers would be more productive. This result can help understand the existence of substantial differences in TFP across countries, even when the technology is common. In particular, Acemoglu and Zilibotti (2001) compare the predictive power of their model in explaining cross-country output differences with that of a comparable neoclassical model, where all countries have access to the same technologies and output is Cobb–Douglas in labor, human and physical capital. Their computations suggest that the proposed mechanism can account for one-third to one half of the total factor productivity gap between the United States and developing countries. Predictions on the pattern of North–South, cross-industry, productivity differences are also tested. Since the South uses the same technology [AL , AH ] as the rest of the world, but it has a higher relative price for skill-intensive goods, it follows that the value of productivity in LDCs relative to that of the North should be higher in skill-intensive sectors. The empirical analysis supports this prediction. The view that countries adopt different technologies out of a world “menu”, and that the choice of the appropriate technology depends on factor endowments, particularly on the average skill of the labor force, finds support in the analysis of Caselli and Coleman (2005). However, these authors also find that many poor countries choose technologies inside the world technology frontier, thereby suggesting that barriers to technology adoption may also be important to explain the low total factor productivity of these countries. 4.3. Trade, inequality and appropriate technology We have seen that directed technical change can help understand inequality, both within and between countries. Several authors have stressed that international trade is another important determinant of income distribution. For example, Wood (1994) argues that the higher competition with imports from LDCs may be responsible for the deterioration in relative wages of low-skill workers in the US in the past decades. Further,
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
141
there is a widespread concern that globalization may be accompanied by a widening of income differences between rich and poor countries. Although the analysis of these issues goes beyond the scope of this paper, we want to argue that R&D-driven endogenous growth models can fruitfully be used to understand some of the links between trade and inequality. In particular, we now show that trade with LDCs can have a profound impact on income distribution, beyond what is suggested by static trade theory, through its effect on the direction of technical change. By changing the relative prices and the location of production, international trade can change the incentives for developing innovations targeted at specific factors or sectors, systematically benefiting certain groups or countries more than others. A key assumption in deriving these results is that, as in the previous paragraph, LDCs do not provide an adequate protection of IPRs. First, consider the effect of trade in the benchmark model of directed technical change. The analysis follows Acemoglu (2002, 2003a). Recall that the profitability of an innovation depends on its market size and the price of the goods it produces, as in Equation (39). What happens to technology if we allow free trade in YL and YH between a skill-abundant North and a skill-scarce South? The market size for innovations does not change, because inventors continue to sell their machines in the North only. But trade, at first, will increases the relative price of skill-intensive goods in the North. To see this, note that trade generates a single world market with a relative price depending on the world supply of goods. Since skills are scarcer in the world economy than in the North alone, trade will increase the relative price of skill-intensive goods in the North (the opposite will happen in the South). In particular, world prices are now given by Equation (41) using world endowments: AH H W −(1−α)/σ PH = . (54) PL AL LW This change in prices, for a given technology, makes skill-complement innovations more profitable and accelerates the creation of skill-complementary machines. Since, along the BG path, both types of innovations must be equally profitable and hence πH = πL , Equation (39) shows that this process continues until the relative price of goods has returned to the pre-trade level in the North. Substituting Equation (54) into (39) and imposing πH = πL , yields the new equilibrium skill bias of technology: LW H N σ AH = W . (55) AL H LN Given that H N /LN > H W /LW , the new technology is more skill-biased and skilled workers in the North earn higher wages. The effect on the skill premium can be seen by substituting (55) into (42): N W N σ −2 H H L wH = . (56) N W wL L H LN The effect of a move from autarky to free trade in the North can be approximated by the elasticity of the skill premium to a change in LW /H W computed at LW /H W =
142
G. Gancia and F. Zilibotti
LN /H N (that is, starting from the pre-trade equilibrium). Equation (56) shows this elasticity to be unity. Thus, if, for example, LW /H W were 4% higher than LN /H N , the model would predict trade to raise the skill premium by the same 4%.32 Without technical change, instead, the reaction of the skill premium to a change in the perceived scarcity of factors due to trade depends on the degree of substitutability of skilled and unskilled workers. From Equation (42), the elasticity of the skill premium to a change in L/H would be 1/σ , less than in the case of endogenous technology as long as σ > 1, i.e., when skilled and unskilled workers are gross substitutes. Therefore, with directed technical change and σ > 1, trade increases the skill premium in the North by more than would otherwise be the case: for example, if the elasticity of substitution is 2, the endogenous reaction of technical progress doubles the impact of trade on wage inequality. Note that another direct channel through which trade can affect factor prices in models of endogenous technical change is by affecting the reward to innovation. If trade increases the reward to innovation (for example, through the scale effect) and the R&D sector is skill-intensive relative to the rest of the economy, trade will naturally spur wage inequality. This mechanism is studied by Dinopoulos and Segerstrom (1999) in a quality-ladder growth model with no scale effects.33 What are the implications of trade opening for cross-country income differences? We have seen that trade induces a higher skill bias in technology; given the result of Acemoglu and Zilibotti (2001) that the excessive skill-complementarity of Northern technologies is a cause of low productivity in Souther countries, it may seem natural to conclude that trade would then increase productivity differences. However, this conclusion would be premature. In the absence of any barriers, trade equalizes the price of goods; given that the production functions adopted so far rule out complete specialization, this immediately implies that factor prices and sectoral productivity are also equalized. This does not mean that trade equalizes income levels; because of their different skill-composition, the North and the South will still have differences in income per capita, but nothing general can be said.34 The fact that trade generates productivity convergence crucially depends on factor prices being equalized by trade. Since factor price equalization is a poor approximation of reality, it is worth exploring the implications of models with endogenous technologies when this property does not hold. A simple way of doing this is to add Ricardian 32 Borjas, Freeman and Katz (1997) show that 4% is a plausible estimate of the increase in the unskilled labor
content of US trade with LDCs between 1980 and 1995. Therefore, this simple exercise may give a sense of how much of the roughly 20% increase in the US skill premium in the same period can be attributed to trade. 33 Recently, other papers have suggested that trade between identical countries may as well increase the skill premium through its effect on technology. See, for example, Epifani and Gancia (2002), Neary (2003) and Thoenig and Verdier (2003). 34 A general result is that the endogenous response of technology makes trade less beneficial for LDCs than would otherwise be the case. This occurs because, after trade opening, the skill premium rises as a result of the induced skill-biased technical change. Given that the North is more skilled-labor abundant, it proportionally benefits more from a higher skill premium.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
143
productivity differences, so that trade opening leads to complete specialization. In this case, the endogenous response of technology to weak IPRs in LDCs becomes a force promoting productivity divergence.35 Further, trade with countries providing weak protection for IPRs may have an adverse effect on the growth rate of the world economy. These results, shown by Gancia (2003), can be obtained by modifying Acemoglu and Zilibotti (2001) as follows. First, we allow the elasticity of substitution between final 1 goods to be larger than one: Y = [ 0 yi (ε−1)/ε di]ε/(ε−1) , with ε > 1. Then, we assume that each good yi can be produced by competitive firms both in the North and the South, using sector-specific intermediates and labor: Ai
S α N 1−α Ai N α S 1−α xi,v dv + ili xi,v dv. yi = (1 − i)li (57) 0
0
There are three important differences with respect to (45). First, (1 − i) and i now capture Ricardian productivity differences between the North and South, implying that the North is relatively more productive in high index sectors. Second, intermediate goods are sector specific, not factor specific (there is now a continuum [0, 1] of technologies, not only two). Third, there is only one type of labor. Given that the endogenous component of technology (Ai ) is still assumed to be common across countries, the sectoral North–South productivity ratio only depends on the Ricardian elements. The new implication is that countries specialize completely under free trade, as each good is only manufactured in the location where it can be produced at a lower cost. The equilibrium can be represented by the intersection of two curves, as in Dornbusch et al. (1977). For any relative wage, the first curve gives the range [0, J ] of goods effiN J = wwS . The second curve combines trade balance ciently produced in the South: 1−J and a BG research arbitrage condition, requiring profits to be equalized across sectors and countries. To find this, the model assumes that the owner of a patent can only extract a fraction λ < 1 of the profits generated by its innovation in the South, so that λ can be interpreted as an index of the strength of international IPRs protection. The trade balance plus the research arbitrage condition turn out to be [see Gancia (2003)]: S 1 σ σ /(1− σ ) di 1− wN J (i) − σ L = λ , (58) σ /(1− σ ) di wS LN J (1 − i) 0 σ > 0 (i.e., ε > 1), the wage gap is with σ ≡ (1 − α)(ε − 1) ∈ (0, 1).36 As long as decreasing in the degree of protection of IPRs in the South, λ. The reason is that weaker protection of IPRs shifts innovations out of Southern sectors and increases the relative N J = wwS , it is easily seen that a weaker productivity of the North. From the condition 1−J 35 The idea that trade may magnify cross-country inequality was put forward by several economists. Some
examples are Stiglitz (1970), Young (1991), Krugman and Venables (1995), Matsuyama (1996), RodriguezClare (1996) and Ventura (1997). 36 σ < 1 guarantees balanced growth across sectors. σ > 0, i.e., an elasticity of substitution between goods greater than one, rules out immiserizing growth.
144
G. Gancia and F. Zilibotti
protection of IPRs in the South, by raising w N /w S , is accompanied by a reduction in sectors [1 − J ] located in the North, because higher wages make the North less competitive. A second result emerges by calculating the growth rate of the world economy. In particular, Gancia (2003) shows the growth rate of the world economy to fall with λ and approach zero if λ is sufficiently low. The reason is that a lower λ shifts innovation towards Northern sectors and, at the same time, induces the relocation of more sectors to the South, where production costs become lower. This, in turn, implies that a wider range of goods becomes subject to weak IPRs and hence, to a low innovation incentive.
5. Complementarity in innovation In the models described so far, innovation has no effect on the profitability of existing intermediate firms. This is a knife-edge property which descends from the specification of the final production technology, (3). In general, however, new technologies can substitute or complement existing technologies. Innovation often causes technological obsolescence of previous technologies. Substitution is emphasized, in an extreme fashion, by Schumpeterian models such as Aghion and Howitt (1992). In such models, innovation provides “better of the same”, i.e., more efficient versions of the pre-existing inputs. Growth is led by a process of creative destruction, whereby innovations do not only generate but also destroy rents over time. This has interesting implications for dynamics: the expectation of future innovations discourages current innovation, since today’s innovators expect a short life of their rents due to rapid obsolescence. More generally, substitution causes a decline in the value of intermediate firms over time, at a speed depending on the rate of innovation in the economy. There are instances, however, where new technologies complement rather than substitute old technologies. The market for a particular technology is often small at the moment of its first introduction. This limits the cash-flow of innovating firms, which initially pose little threat to more established technologies. However, the development of new compatible applications expands the market for successful new technologies over time, thereby increasing the profits earned by their producers. Rosenberg (1976) discusses a number of historical examples, where such complementarities were important. A classical example is the steam engine. This had been invented in the early part of the XVIIIth Century, but its diffusion remained very sporadic before a number of complementary innovations (e.g., Watt’s separate condenser) made it competitive with the waterwheels, which remained widespread until late in the XIXth Century. Complementarity in innovation raises interesting issues concerning the enforcement and design of intellectual property rights. For instance, what division of the surplus between basic and secondary innovation maximizes social welfare? This issue is addressed by Scotchmer and Green (1995) who construct a model where innovations are sequentially introduced, and the profits of major innovators can be undermined by subsequent derivative innovations. In this case, the threat of derivative innovations can reduce the
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
145
incentive for firms to invest in major improvements in the first place. However, too strong a defense of the property right of basic innovators may reduce the incentive to invest in socially valuable derivative innovations. Scotchmer and Green (1995) show that the optimal policy in fact consists of a combination of finite breadth and length of patents. Scotchmer (1996) instead argues that it may be optimal to deny patentability to derivative innovations, instead allowing derivative innovations to be developed under licensing agreements with the owner of the basic technology. More recently, Bessen and Maskin (2002) show that when there is sufficient complementarity between innovations (as in the case of the software industry), weak patent laws may be conducive to more innovation than strong patent laws. The reason is that while the incumbent’s current profit is increased by strong patent laws, its prospect of developing future profitable innovation is reduced when patent laws inhibit complementary innovations. While this literature focuses on the partial equilibrium analysis of single industries, complementarity in innovation also has implications on broader development questions. Multiple equilibria originating from coordination failures [of the type emphasized, in different contexts, by Murphy, Shleifer and Vishny (1989), and Cooper and John (1988)] can arise when there is complementarity in innovation. Countries can get locked-in into an equilibrium with no technology adoption, and temporary big-push policies targeting incentives to adopt new technologies may turn out to be useful.37 One such example is Ciccone and Matsuyama (1996). In their model, multiple equilibria and poverty traps may arise from the two-way causality between the market size of each intermediate good and their variety: when the availability of intermediates is limited, final good producers are forced to use a labor intensive technology which, in turn, reduces the incentive to introduce new intermediates. Young (1993) constructs a model where innovation expands the variety of both intermediate and final goods. New intermediate inputs are not used by mature final industries, and their market is initially thin. The expansion of the market for technologies over time creates complementarity in innovation. The details of this model are discussed in the remainder of this section. To this aim, we augment the benchmark model of Section 2 with the endogenous expansion in the variety of final goods.38 Over time, innovative investments make new intermediate inputs available to final producers, as in Romer’s model. However, as a by-product (spillover), they also generate an equivalent expansion of the set of final goods that can be produced. There are no property rights defined on the production of new final goods, and these are produced by competitive firms extraneous to the innovation process.
37 Interestingly, in models with complementarity in innovation, market economies may be stuck in no-growth
traps that are inefficient in the sense that the optimal intertemporal allocation would require positive investment and growth. See, for example, Ciccone and Matsuyama (1999). 38 Models featuring an expanding variety of final products include Judd (1985), Grossman and Helpman (1989, Chapter 3) and, more recently, Xie (1998) and Funke and Strulik (2000). Here, we follow Young (1993) which, in turn, is close to Judd’s paper.
146
G. Gancia and F. Zilibotti
At will now denote the measure of both final goods and intermediate goods available in the economy at t. Final products are imperfect substitutes in consumption, and the instantaneous utility function is:39 At Vt = ln(Cs,t ) ds, 0
with total utility being ∞ U= e−ρt Vt dt. 0
This specification implies that consumers’ needs grow as new goods become available. Suppose, for instance, that a measure ε of new goods is introduced between time t and t + j . At time t, consumers are satisfied with not consuming the varieties yet to be invented. However, at time t + j , the same consumers’ utility would fall to minus infinity if they did not consume the new goods. The productive technology for the s’th final good is given by min[sΘ,A] 1/α α Qs = (59) xj,s dj , 0
where Θ 1 is a parameter. Note that labor is not used in the final goods production. First, to build the intuition in the simplest case, we maintain that all final goods are produced with the same technology employing all available varieties of intermediate inputs. More formally, we characterize the equilibrium in the limit case where Θ → ∞, so that min[sΘ, A] = A. This assumption will be relaxed later. We use the “labor-for-intermediates” model introduced in Section 2.2, where labor is used for research and intermediate production and the productivity of labor in intermediate production equals At . We choose the nominal wage as the numeraire.40 Hence, the profit of an intermediate producer can be expressed as: 1−α x 1 − α L − Lx (60) = . α A α A Note that profits fall over time at the rate at which knowledge grows. In a BG equilibrium, the interest rate is constant and A grows at the constant rate γ . Free entry implies: ∞ 1 1 − α L − γ /δ e−rτ πτ dτ = , α A (r + γ ) δA t t t π=
39 This is the benchmark specification in Young (1993), where it is then extended to general CES preferences
across goods. The logarithmic specification is analytically convenient because of the property that consumers spend an equal income share on all existing goods. 40 Note that we cannot simply set the price of the final good as the numeraire, as there is an increasing variety of final goods.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
147
where we have used the fact that γ = δLx . Simplifying terms yields 1 − α δL − γ 1 α r +γ
(61)
The intertemporal optimality condition for consumption also differs from the benchmark model. In particular, if E denotes the total expenditure in final goods, the Euler condition is:41 N˙ t E˙ =r −ρ+ . E Nt In a BG equilibrium, the total expenditure on consumption goods is constant. Hence, r = ρ − γ.
(62)
This expression can be substituted into (61) to give a unique solution for γ . As long as growth is positive, we have αρ , γ = δL − 1−α which is almost identical to (15), except for the constant term α/(1 − α) being replaced by 1/α. In the limit case considered so far (Θ → ∞), the model is isomorphic to Romer (1990). Next, we move to the general case where Θ 1 is finite. This implies that final producers cannot use the entire range of intermediate goods. In particular, an intermediate good indexed by s cannot be used by “mature” final industries having an index j , such that j < s/Θ. This assumption captures the idea that a technology mismatch develops over time between mature final good industries and new technologies.42 An important implication of this assumption is that, when introduced, a new technology (intermediate input) is only required by a limited number of final industries. Thus, the monopolist producing a new variety has a small cash-flow. This is especially true when the parameter Θ is small: as Θ → 1, there is no demand for a new intermediate good at the time of its first appearance. However, the market for technologies expands over time, as new final goods using “modern” technologies appear. This dynamic market size effect generates complementarity in the innovation process. An innovator is eager to see rapid technical progress, as this expands the number of users of the new technology. Countering this effect, there is a process of “expenditure diversion” that reduces, ceteris paribus, the demand for each intermediate good. Over time, technical progress expands the number of intermediate inputs over which final producers spread their 41 See Young (1993, p. 783) for the derivation of this Euler equation. 42 In principle, it would seem natural to assume that new final goods do not use very old intermediate goods.
Young (1993, p. 780) argues that allowing for this possibility would not change the main results, but would make the analysis more involved.
148
G. Gancia and F. Zilibotti
demand. As noted above, the total expenditure on final goods is constant in a BG equilibrium. Since final good firms make zero profits at all times, and intermediates are the only inputs, the total expenditure on the intermediate goods must also be constant. Therefore, an increase in A dilutes the expenditure over a larger mass of intermediate goods, and reduces the profit of each existing intermediate firm. This effect generates substitution rather than complementarity in innovation. The dynamic market size effect may dominate for young intermediate firms. But as a technology becomes more mature, the expenditure dilution effect takes over. Thus, firms can go through a life-cycle: their profit flow increases over time at an earlier stage and decreases at a later stage. We denote by π(Aτ , At ) the profit realized at time τ by an intermediate producer who entered the market in period t < τ . Solving the profit maximization problem for the intermediate monopolist, subject to the demand from final industries, leads to the following expression: 1−α γ (τ − t) − 1 π(Aτ , At ) = (63) (L − Lx ) 1 + . αAτ Θ It is easily verified that as Θ → ∞, the solution becomes identical to (60), where nominal profits fall at the same rate as At . Free-entry implies: ∞ 1 (64) e−r(τ −t) π(Aτ , At ) dτ . δAt t Solving the integral on the left-hand side, using the Euler condition, r + γ = ρ, and simplifying terms yields the following equilibrium condition: fFE (γ ) =
1−α (δL − γ ) γ + ρ(Θ − 1) 1, αρ 2 Θ
(65)
where all terms but γ are parameters. For sufficiently large values of Θ, i.e., when (γ ) < 0 and the equilibrium is unique. the market for new technology is large, fFE However, if Θ < 1 + δL/ρ, fFE (γ ) is non-monotonic, and multiple equilibria are possible. Figure 2 describes the three possible cases. As long as ρ > γ , which is a necessary and sufficient condition for the interest rate to be positive, fFE (γ ) is increasing in Θ. For a range of small Θ’s, there is no equilibrium with positive innovation (lower curve). The only equilibrium is a point such as X , featuring zero growth. For an intermediate range of Θ, we have fFE (γ ) = 1 in correspondence of two values of γ (intermediate curve). This implies that (for generic economies), there exist three equilibria, where equilibria such as point X feature zero innovation and growth. Firms contemplating entry expect no expansion of the market size for new technologies. Furthermore, such market size is too small to warrant profitable deviations, and the expectation of no innovation is fulfilled in equilibrium. Equilibria such as point Y are
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
149
Figure 2.
characterized by local complementarity in innovation: the expectation of higher future innovation and growth increases the value of new firms, stimulating current entry and innovation. In steady-state (BG), this implies a positive slope of the locus fFE (γ ).43 Eventually, for sufficiently high growth rates, the diversion effect dominates. Thus, in an equilibrium like Z, the value of innovating firms depends negatively on the speed of innovation.44 Finally, for a range of large Θ’s, substitution dominates throughout (upper curve). The initial market for new technologies is sufficiently large to make the expenditure diversion effect dominate the market size effect, even at low growth rates. The equilibrium is unique, and the solution is isomorphic to that of the benchmark model of expanding variety.
43 As mentioned above, firms go through a life-cycle here. When a new technology is introduced, the profit
flow of an innovating firm is small. As time goes by, the expenditure diversion effect becomes relatively more important. The value of a firm upon entry is the PDV of its profit stream. Local complementarity occurs if, for a particular γ , profits increase at a sufficiently steep rate in the earlier part of the firm’s life-cycle. 44 If the expectational stability of the equilibria in the sense of Evans and Honkapohja (2001) is tested, equilibria such as point Y are not found to be E-stable, while equilibria such as X and Z are stable. See the discussion in Section 7.2.
150
G. Gancia and F. Zilibotti
6. Financial development A natural way in which the expansion of the variety of industries can generate complementarities in the growth process is through its effects on financial markets. Acemoglu and Zilibotti (1997) construct a model where the introduction of new securities, associated with the development of new intermediate industries, improves the diversification opportunities available to investors. Investors react by supplying more funds, which fosters further industrial and financial development, generating a feedback.45 The model offers a theory of development. At early stages of development, a limited number of intermediate industries are active (due to technological nonconvexities), which limits the degree of risk-spreading that the economy can achieve. To avoid highly risky investments, agents choose inferior but safer technologies. The inability to diversify idiosyncratic risks introduces a large amount of uncertainty in the growth process. In equilibrium, development proceeds in stages. First, there is a period of “primitive accumulation” with a highly variable output, followed by take-off and financial deepening and finally, steady growth. Multiple equilibria and poverty traps are possible in a generalized version of the model. The theory can explain why the growth process is both slow and highly volatile at early stages of development, and stabilizes as an economy grows richer. Evidence of this pattern can be found in the accounts of pre-industrial growth given by a number of historians, such as Braudel (1979), North and Thomas (1973) and De Vries (1990). For instance, in cities such as Florence, Genoa and Amsterdam, prolonged periods of prosperity and growth have come to an end after episodes of financial crises. Interestingly, these large set-backs were not followed (as a neoclassical growth model would instead predict) by a fast recovery but, rather, by long periods of stagnation. Similar phenomena are observed in the contemporary world. Acemoglu and Zilibotti (1997) document robust evidence of increases in GDP per capita being associated with large decreases in the volatility of the growth process. It has also been documented that higher volatility in GDP is associated with lower growth [Ramey and Ramey (1995)]. We here describe a simplified version of the model. Time is discrete. The economy is populated by overlapping generations of two-period lived households. The population is constant, and each cohort has a mass equal to L. There is uncertainty in the economy, which we represent by a continuum of equally likely states s ∈ [0, 1]. Agents are assumed to consume only in the second period of their lives.46 Their preferences are parameterized by the following (expected) utility function, inducing unit relative risk
45 This paper is part of a recent literature on the two-way relationship between financial development and
growth. This includes Bencivenga and Smith (1991), Greenwood and Jovanovic (1990) and Zilibotti (1994). In none of these other papers does financial development take the form of an expansion in the “variety” of assets. 46 This is for simplicity. Acemoglu and Zilibotti (1997) assume that agents consume in both periods. It is also possible to study the case of a general CRRA utility function.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
aversion:
Et U (ct+1 ) =
151
1
log(cs,t+1 )ds.
(66)
0
The production side of the economy consists of a unique final good sector, and a continuum of intermediate industries. The final good sector uses intermediate inputs and labor to produce final output. Output in state s ∈ [0, 1] is given by the following production function: Ys,t = (xs,t−1 + xΦ,s,t−1 )α L1−α .
(67)
For simplicity, we assume L = 1. The term in brackets is “capital”, and it is either produced by a continuum of intermediate industries, each producing some state-contingent amount of output (xs ), or a separate sector using a “safe technology” (xΦ ). The measure of the industries with a state-contingent production, At , is determined in equilibrium, and At can expand over time, like in Romer’s model, but it can also fall. Moreover, At ∈ [0, 1], i.e., the set of inputs is bounded. In their youth, agents work in the final sector and earn a competitive wage, ws,t = (1 − α)Ys,t . At the end of this period, they take portfolio decisions: they can place their savings in a set of risky securities ({Fi }i∈[0,At ] ), consisting of state-contingent claims to the output of the intermediate industries or, in a safe asset (φ), consisting of claims to the output of the safe technology. After the investment decisions, the uncertainty unravels, the security yields its return and the amount of capital brought forward to the next period is determined. The capital is then sold to final sector firms and fully depreciates after use. Old agents consume their capital income and die. Intermediate industries use final output for production. An intermediate industry i ∈ [0, At ] is assumed to produce a positive output only if state s = i occurs. In all other states of nature, the firm is not productive. Moreover, the ith industry is only productive if it uses a minimum amount of final output, Mi , where
D Mi = max 0, (i − x) , (1 − x) with x ∈ (0, 1). This implies that some intermediate industries require a certain minimum size, Mi , before being productive. In particular, industries i x have no minimum size requirement, and for the rest of the industries, the minimum size requirement increases linearly with the index i. To summarize, the intermediate technology is described by the following production function: RFi if i = s and Fi Mi , xi,s = 0 otherwise. Since there are no start-up costs, all markets are competitive. Thus, firms retain no profits, and the product is entirely distributed to the holders of the securities. The j th security entitles its owner to a claim to R units of capital in state j (as long as the minimum size constraint is satisfied, which is always the case in equilibrium), and otherwise
152
G. Gancia and F. Zilibotti
to nothing. Savings invested in the “safe technology” give the return xΦ,s = rφ,
∀s ∈ [0, 1],
where r < R. Thus, one unit of the safe asset is a claim to r units of capital in all states of nature. Since the risky securities yield symmetric returns, and there is safety in numbers, it is optimal for risk-averse agents to hold a portfolio containing all available securities in equal amounts. More formally, the optimal portfolio decision features Fi = F , for all i ∈ [0, At ]. We refer to this portfolio consisting of an equal amount of all traded risky securities as a balanced portfolio. If At = 1, a balanced portfolio of risky securities bears no risk, and first-order dominates the safe investment. However, due to the presence of technological nonconvexities (minimum size requirements), not all industries are in general activated. When At < 1, the inferior technology is safer, and there is a trade-off between risk and productivity. In this case, the optimal investment decision of the representative saver can be written as:
max At log ρG,t+1 (RFt + rφt ) + (1 − At ) log ρB,t+1 (rφt ) , (68) φt,Ft
subject to φt + At Ft wt .
(69)
ρs˜,t+1 denotes the rate of return of capital, which is taken as parametric by agents, and does not affect the solution of the program.47 Agents also take At , i.e., the set of securities offered, as parametric. Simple maximization yields: (1 − At )R wt , R − rAt
R−r = F (At ) ≡ R−rAt wt , 0
φt∗ = ∗ Fi,t
(70) ∀i At , ∀i > At .
(71)
Figure 3 expresses the demand for each risky asset, F (At ) (FF schedule), as a function of the measure of intermediate industries which are active. The FF schedule is 47 In equilibrium:
ρG,t+1 = α(RFt + rφt )α−1 and ρB,t+1 = α(rφt )α−1 . ρG,t+1 applies in the “good state”, i.e., when the realized state is i At , while ρB,t+1 is the marginal product of capital in the “bad” state, when the realized state is i > At and no risky investment pays off.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
153
Figure 3.
upward sloping, implying that there is complementarity in the demand for risky assets: the demand for each asset grows with the variety of intermediate industries. Complementarity arises because the more active are intermediate industries, the better is risk-diversification. Thus, as At increases, savers shift their investments away of the safe asset into high-productivity risky projects (the “stock market”). Such complementarity hinges on risk aversion being sufficiently high.48 In general, similar to Young (1993), an increase in A creates two effects. On the one hand, investments in the stock market become safer because of better diversification opportunities, which induces complementarity. On the other hand, investments are spread over a larger number of assets, inducing substitution. With sufficiently high risk aversion, including the unit CRRA specification upon which we focus, the first effect dominates. The equilibrium measure of active industries, A∗t , is determined (as long as A∗ < 1) by the following condition: F A∗t = MA∗t .
48 Suppose agents were risk averse, but only moderately so. Suppose, in particular, that they were so little
risk-averse that they would decide not to hold any safe asset in their portfolio. Then, an expansion in the set of risky securities would induce agents to spread their savings (whose total amount is predetermined) over a larger number of assets. In this case, assets would be substitutes rather than complements.
154
G. Gancia and F. Zilibotti
In Figure 3, the equilibrium is given by the intersection of schedules FF and MM, where the latter represents the distribution of minimum size requirements across industries. Intuitively, A∗t is the largest number of industries for which the technological non-convexity can be overcome, subject to the demand of securities being given by (71).49 Growth increases wage income and the stock of savings over time. In equilibrium, this induces an expansion of the intermediate industries, A∗t . This can once more be seen in Figure 3: growth creates an upward shift of the FF schedule, causing the equilibrium to move to the left. Therefore, growth triggers financial development. In particular, when the stock of savings becomes sufficiently large, the financial market is sufficiently thick to allow all industries to be active. In the case described by the dashed curve, FF , the economy is sufficiently rich to afford A∗t = 1. The inferior safe technology is then abandoned. Financial development, speeds up growth by channelling investments towards the more productive technology. The stochastic equilibrium dynamics of GDP can be explicitly derived: α ∗ F (Y ) = (1 − α) r(1 − At ) RY prob. 1 − A∗t , B t t ∗ Yt+1 = (72) R − rA t α prob. A∗t , FG (Yt ) = (1 − α)RYt where A∗t = A(Yt−1 ) 1 is the equilibrium measure of intermediate industries, such that A 0.50 The first line corresponds to the case of a “bad realization” at time t, such that s ∈ (A∗t , 1]. In this case, none of the active intermediate industries turned out to pay-off at time t, and capital at time t + 1 is only given by the return of the safe technology. The second line corresponds to the case of a “good realization” at t, such that s ∈ [0, A∗t ]. In this case, the risky investment paid off at time t, and capital and output are relatively large at time t + 1. Note that the probability of a good realization increases with the level of development, since A (Yt−1 ) 0 (with strict inequality as long as A∗ < 1). Figure 4 describes the dynamics. The two schedules represent output at time t + 1 as a function of output at time t conditional on good news (FG (Yt )) and bad news (FB (Yt )), respectively. At low levels of capital (Y YL ), the marginal product of capital is very high, which guarantees that growth is positive, even conditional on bad news. In the intermediate range where Y ∈ [YL , YM ], growth only occurs if news is good, since FB (Yt ) < Yt < FG (Yt ). The threshold YL is not a steady-state; however, it is a point around which the economy will spend some time. When the initial output is below YL , the economy necessarily grows towards it. When it is above YL , output falls back whenever bad news occurs. So, in this region, the economy is still exposed to undiversified risks, and experiences fluctuations and set-backs. Finally, for Y YM , there are 49 Acemoglu and Zilibotti (1997) show the laissez-faire portfolio investment to be inefficient. Efficiency
would require more funds to be directed to industries with large non-convexities, i.e., agents not holding a balanced portfolio. The inefficiency is robust to the introduction of a rich set of financial institutions. 50 Acemoglu and Zilibotti (1997) derive a closed-form solution for A∗ that we do not report here. t
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
155
Figure 4.
enough savings in the economy to overcome all technological non-convexities. When the economy enters this region, all idiosyncratic risks are removed, and the economy deterministically converges to YH .51 Note that it may appear as if, in the initial stage, countries striving to take off do not grow at a sustained rate during long periods. The demand for insurance takes the form of investments in low-productivity technologies, and poor economies tend to have low total factor productivity and slow growth. In the case described by Figure 4, the economies “almost surely” converge to a unique steady-state. Different specifications of the model can, however, lead to less optimistic predictions. With higher risk aversion, for instance, traps can emerge, as in the example described in Figure 5. An economy starting with a GDP in the region [0, YMM ) would never attain the high steady-state YH , and would instead perpetually wander in the trapping region [0, YLL ]. Conversely, an economy starting above YM would certainly converge to the high steady-state, YH . Finally, the long-run fate of an economy starting in the region [YMM , YM ] would be determined by luck: an initial set of positive
51 That the economy converges “almost surely” to a steady-state where all risk is diversified away only occurs under parameter restrictions ensuring that Y SS > Y 1 . Although the model presented here is neoclassical and
features zero growth in the long run, it is possible to augment it with spillover of the learning-by-doing type, as in Romer (1986), and make it generate self-sustained growth.
156
G. Gancia and F. Zilibotti
Figure 5.
draws would bring this economy into the basin of attraction of the good equilibrium. A single set-back, however, would forever jeopardize its future development.52 The model can be extended in a number of directions. A two-country extension shows that international capital flows may lead to divergence, rather than convergence between economies. This result is due to the interplay between two forces: first, decreasing returns to capital would tend to direct foreign investments towards poorer countries, as in standard neoclassical models. Second, the desire to achieve better diversification pushes investments towards thicker markets. The latter force tends to prevail at some earlier stages of the development process. So, poor countries suffer an outflow of capital, which spills over to lower income and wages for the next generation, thereby slowing down the growth process. The analysis of capital flows, financial integration and financial crises in the context of similar models is further developed in recent papers by Martin and Rey (2000, 2001 and 2002). A different extension of the model is pursued by Cetorelli (2002) who shows that the theory can account for phenomena such as “club convergence”, economic miracles, growth disasters and reversals of fortune.
52 Consider, for instance, the limit case where agents are infinitely risk-averse. In this case, agents refuse
to invest in the stock-market as long as this entails some uncertainty, i.e., as long as there are not enough savings in the economy to open all industries. Thus, an economy starting above YM converges to YH , while an economy starting below YM converges to YL , and is stuck in a poverty trap.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
157
Recent empirical studies analyze implications of the theory about the patterns of risk-sharing and diversification. Kalemli-Ozcan et al. (2001, 2003) document that regions with access to better insurance through capital markets can afford a higher degree of specialization. Using cross-country data at different levels of disaggregation, Imbs and Wacziarg (2003) find robust evidence of sectoral diversification increasing in GDP. However, their findings also suggest that, at a relatively late stage of the development process, the pattern reverts and countries once more start to specialize. This tendency for advanced countries to become more specialized as they grow can be explained by factors emphasized by the “new economic geography” literature [Krugman (1991)], from which the theory described in this section abstracts, such as agglomeration externalities and falling transportation costs.
7. Endogenous fluctuations In the models reviewed so far, the economies converge in the long run to balancedgrowth equilibria characterized by linear dynamics. Growth models with expanding variety and technological complementarities can, however, generate richer long-run dynamics, including limit cycles. In this section, we review two such models. In the former, based on Matsuyama (1999), cycles in innovation and growth arise from the deterministic dynamics of two-sector models with an endogenous market structure. The theory can explain some empirical observations about low-frequency cycles, and their interplay with the growth process. In particular, it predicts that waves of rapid growth mainly driven by “factor accumulation” are followed by spells of innovationdriven growth. Interestingly, these latter periods are characterized by lower investments and slower growth. This is consistent with the findings of Young (1995) that the growth performance of East-Asian countries was mainly due to physical and human capital accumulation, while there was little total factor productivity (TFP) growth. According to Matsuyama’s theory, the observation of low TFP growth should not lead to the pessimistic conclusions that growth is destined to die-off. Rather, rapid factor accumulation could set the stage for a new phase of growth characterized by more innovative activity. The predictions of this theory bear similarities to those of models with General Purpose Technologies (GPT), e.g., Helpman (1998) and Aghion and Howitt (1998, Chapter 8). For instance, they predict that a period of rapid transformation and intense innovation (e.g., the 1970’s) can be associated with productivity slowdowns. However, GPT-based theories rely on the exogenous arrival of new “fundamental” innovations generating downstream complementarities. In contrast, cycles in Matsuyama (1999) are entirely endogenous.53
53 Cyclical equilibria can also emerge in Schumpeterian models, due to the dynamic relationship between
innovative investments and creative destruction. An example is the seminal contribution of Aghion and Howitt (1992). More recently, Francois and Lloyd-Ellis (2003) construct a Schumpeterian model where entrepreneurs
158
G. Gancia and F. Zilibotti
In the latter model, based on Evans et al. (1998), cycles in innovation and growth are instead driven by expectational indeterminacy. The mechanism in this paper is different, as cycles hinge on multiple equilibria and sunspots. Some main predictions are also different: contrary to Matsuyama (1999), the equilibrium features a positive comovement of investments and innovation. The main contribution of the paper is to show that cycles can be learned by unsophisticated agents holding adaptive expectations. Thus, the predictive power of the theory does not rest on the assumption that agents’ expectations are rational and that agents can compute complicated dynamic equilibria. 7.1. Deterministic cycles Matsuyama (1999) presents a model of expanding variety where an economy can perpetually oscillate in equilibrium between periods of innovation and periods of no innovation. Cycles arise from the deterministic periodic oscillations of two state variables (physical capital and knowledge). Unlike the model that will be discussed in the next section, the equilibrium is determinate and there are no multiple steady-states. More specifically, the source of the oscillatory dynamics is the market structure of the intermediate goods market. Monopoly power is assumed to be eroded after one period. The loss of monopoly power is due to the activity of a competitive fringe which can copy the technology with a one-period lag. In every period, new industries are monopolized, while mature industries are competitive. The profits of innovators depend on the market structure of the intermediate sector. The larger is the share of competitive industries in the intermediate sector, the lower is the profit of innovative firms, since competitive industries sell larger quantities and charge lower prices. In periods of high innovation, a large share of industries are monopolized, which increases the profitability of innovation, thereby generating a feedback. In these times, investment in physical capital is low due to the crowding out from the research activity. Conversely, times of low innovation are times of high competition, since old monopolies lose power and there are few new firms. Thus, the rents accruing to innovative firms are small. In these periods, savings are invested in physical capital, and while innovation is low, the high accumulation of physical capital creates the conditions for future innovation to be profitable. Time is discrete. The production of final goods is as in (3), where we set Ly = L = 1. Intermediate goods are produced using physical capital, with one unit of capital producing one unit of intermediate product, x. Innovation also requires capital, with a requirement of µ units of capital per innovation. Monopoly power is assumed to last one period only. Therefore, in period t, all intermediate inputs with an index z ∈ [0, At−1 ]
can decide to time the implementation of innovations [similarly to Shleifer (1986)]. In this model, agents time the implementation so as to profit from buoyant demand and maximize the duration of their leadership. This mechanism leads to a clustering of innovations and endogenous cycles. While this model can explain some features of fluctuations at business cycle frequencies, Matsuyama’s model is better suited for the analysis of long waves.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
159
are competitively priced, whereas all those with an index z ∈ [At−1 , At ] are monopolistically priced. The prices of competitive and non-competitive varieties are ptc = rt and ptm = rt /α, respectively, where the superscript h ∈ {c, m} denotes the market structure. The relative demand for two varieties xtc and xtm must be xtc = xtm
ptc ptm
−1/(1−α)
= α −1/(1−α) .
(73)
The one-period monopoly profit is πt = ptm xtm − rt xtm = xtm rt (1 − α)/α. Since patents expire after one period, πt is also the value of a monopolistic firm at the beginning of period t. Therefore, free-entry implies: 1−α m (74) xt µ, α with equality holding when innovation is positive. Capital is assumed to fully depreciate after each period. The stock of capital can be allocated to research or intermediate production, subject to the following resource constraint: Kt−1 = At−1 xtc + (At − At−1 ) xtm + µ , implying At − At−1
(1 − α)Kt−1 −α/(1−α) , = At−1 max 0, −α µAt−1
(75)
where we have used (73) and (74) to eliminate xtc and xtm . As shown by (75), there exists a threshold to the capital-knowledge ratio that triggers positive innovation. In particular, innovation occurs if Kt−1 /At−1 α −α/(1−α) (1 − α)−1 µ ≡ kL . If Kt−1 /At−1 < kL , then, all capital is allocated to intermediate production, all intermediate industries are competitive and final production is given by the standard neoclassical Cobb–Douglas technology: 1−α α Yt = At−1 Kt−1 .
(76)
In this case, an economy is said to be in a “Solow regime”, with decreasing returns to capital. Since there is no investment in innovation, A is constant and the dynamics has a neoclassical character. In contrast, if Kt−1 /At−1 kL , then a positive share of the capital stock is allocated to innovation and final production equals: α α α −1/(1−α) α µ + (At − At−1 ) µ . Yt = At−1 α 1−α 1−α Using (75) and simplifying terms, this equation can be written as Yt = DKt−1 ,
(77)
160
G. Gancia and F. Zilibotti
where D ≡ (kL )−(1−α) . In this case, the returns to capital are constant, like in endogenous growth models, and the economy is said to be in a “Romer regime”.54 For tractability, we assume a constant savings rate, implying that Kt = sYt .55 Define kt = kL−1 · Kt /At as the (adjusted) capital-to-knowledge ratio. Then, standard algebra using (75), (76) and (77) establishes the following equilibrium law of motion: α if kt−1 < 1, sDkt−1 sDkt−1 kt = f (kt−1 ) = (78) if kt−1 1. −α/(1−α) 1+α (kt−1 − 1) The mapping kt = f (kt−1 ) has two fixed points. The first is k = 0, the second can either be k = (sD)1/(1−α) ≡ kˆ1 , if sD 1, or k = 1 + α α/(1−α) (sD − 1) ≡ kˆ2 , if sD 1. In the former case, the fixed point lies in the range of the “ Solow regime”, while in the latter, it lies in the range of the “Romer regime”.56 Three cases are possible: 1. If sD 1, the economy converges monotonically to k = kˆ1 . In this case, the economy never leaves the Solow regime, and there are no innovative investments. The neoclassical dynamics converge to a stagnating level of GDP per capita. 2. If sD > max{1, α −α/(1−α) − 1}, then capital first monotonically accumulates in the Solow regime, with no innovation. The economy overcomes the development threshold, k = 1 in finite time, and the process of innovation starts thereafter. Eventually, the economy converges to the BG equilibrium kˆ2 in an oscillatory fashion. In the BG, capital and knowledge are accumulated at the same positive rate, and income per capita grows over time. 3. If sD ∈ (1, α −α/(1−α) − 1], the economy does not converge asymptotically to any BG equilibrium, and perpetually oscillates in the long run between the Solowand the Romer-regime. This case is described by Figure 6. On the one hand, there is no steady-state in the Solow-regime, which rules out that the economy can be trapped in a stable equilibrium with no innovation. On the other hand, the steadystate kˆ2 is locally unstable and cannot be an attractor of the dynamics in itself. Instead, there exists a period-2 cycle, such that one of the periodic points lies in the Solow regime (kS ), while the other lies in the Romer regime (kR ).57 The period-2 54 Zilibotti (1995) finds similar dichotomic equilibrium dynamics in a one-sector model with learning-by-
doing spillovers. Economies may converge to a stationary steady-state with “Solow dynamics” or embark on a virtuous path of “Romer dynamics” with self-sustained growth. Cycles cannot arise in equilibrium, while multiple self-fulfilling prophecies exist. 55 Matsuyama (2001) relaxes this restriction and characterizes equilibrium by a second-order difference equation. Some of the main results, like the existence of a period-2 cycle, survive this generalization. 56 It is easily verified that f (0) > 1, f (kˆ ) = α ∈ (0, 1), and f (kˆ ) = −(α −α/(1−α) − 1)/(sD), where 1 2 f (kˆ2 ) ∈ (−1, 0) if sD > α −α/(1−α) − 1 and f (kˆ2 ) < −1 if sD ∈ (1, α −α/(1−α) − 1). These properties are used to establish the results discussed below. 57 A period-2 cycle exists if, given a mapping x t+1 = f (xt ), f (f (·)) has fixed points other than the fixed point of f (·). A sufficient condition is that (i) f (·) is continuous, (ii) there exists a closed, finite interval, I , such that f (I ) ⊂ I and (iii) f (·) has an unstable fixed point. (i) and (iii) are clearly satisfied; (iii) is established in the next footnote for the interval Iabs .
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
161
Figure 6.
cycle is not necessarily stable, and if it is unstable, the economy can converge to cycles of higher periodicity or feature chaos. A general property of the dynamics is that the economy necessarily enters the region Iabs = [f (f (1)), f (1)] (shaded in Figure 6), and never escapes from it.58 In the case described by Figure 6, the model predicts that a poor economy would first grow through capital accumulation, and eventually enter the absorbing region Iabs . Then, there is an alternation of periods of innovation and periods of no innovation. GDP per capita grows on average, but at a non-steady rate, and there are cycles in the innovative activity. Interestingly, output and capital grow more quickly in periods of no innovation (Solow regime) than in periods of high innovation (Romer regime). Another implication is that if an economy grows quickly, but has a low TFP growth, this does not imply that growth will die-off. Rather, fast capital accumulation can create the conditions for future waves of innovation, and vice versa.
58 To prove this result, two properties of the mapping need to be shown. First, f (·) must be unimodal, i.e.,
(i) f (·) must be continuous; (ii) f (·) must be increasing in some left-hand neighborhood of 1 and decreasing in some right-hand neighborhood of 1. Second, it must be the case that f (f (1)) < 1. That f is unimodal is immediate by inspection. After some algebra, it can also be proved that f (f (1)) < 1.
162
G. Gancia and F. Zilibotti
7.2. Learning and sunspots Evans et al. (1998) propose the following generalization of the technology (3) for final production: A φ ζ 1−α xj dj , Y =L (79) 0
where ζ φ = α. This specification encompasses the technology (3), in the case of φ = 1, and allows intermediate inputs to be complements or substitutes. They focus on the case of complementarity (φ > 1), and show that in this case, the equilibrium can feature multiple steady states, expectational indeterminacy and sunspots. They emphasize the possibility of equilibria where the economy can switch stochastically between periods of high and low growth. Time is discrete, and intermediate firms rent physical capital from consumers to produce intermediate goods. One unit of capital is required per unit of intermediate good produced. Capital is assumed not to depreciate. The resource constraint of this economy is: Kt+1 − Kt , Yt = Ct + Kt · χ Kt where χ(.) is a function such that χ > 0, χ 0. If there are no costs of adjustment, then, χ(x) = x. If χ > 0, there are convex costs of adjustments. By proceeding as in Section 2, we can characterize the equilibrium of the intermediate industry.59 The profit of intermediate producers, in particular, turns out to be: π = ΩAξ (r pK )α/(α−1) ,
(80)
where ξ ≡ (φ − 1)/(1 − α) and Ω ≡ (1 − are two positive constant. We denote by pK the relative price of capital, expressed in terms of the consumption good numeraire. If there are no adjustment costs, then, pK = 1 while, in general, pK = χ (.). Note that profits increase with A, as long as φ > 1. Two technical assumptions ensure that the model has BG properties. First, the design ξ of a new good requires At units of output. Second, innovative investments incurred at t only give the first profit in period t + 1. Free entry then implies: ζ )ζ (1+α)/(1−α) φ 1/(1−α) L
∞ s=0
πt+s ξ pK,t At . (1 + r)s+1
(81)
In a BG equilibrium, consumption and capital grow at the common rate, γ . When φ > 1, this rate exceeds the growth rate of technical knowledge, γA ≡ At+1 /At . In particular, 59 Note that firms rent, and do not own, their capital stock. Adjustment costs are borne at the aggregate level,
not at the level of each decision unit. Therefore, it continues to be legitimate to write the profit maximization problem for intermediate producers as a sequence of static maximization problems.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
163
Figure 7. 1+ξ
it can be shown that γ = γA . Substituting (80) into (81), and solving, yields: (φ−α)/(φ−1) , γ = 1 + r − Ω(pK )−1/(1−α) r −α/(1−α)
(82)
which is the analogue of Equation (13) in the benchmark model. The model is closed by the (discrete time) Euler equation for consumption:
1/σ γ = β(1 + r) ,
(83)
where β is the discount factor. Equations (82) and (83) fully characterize the equilibrium. Figure 7 provides a geometric representation for the case of logarithmic preferences and zero adjustment cost (pK = 1). The SS curve is linear, with the slope β −1 . The DD curve is also positively sloped. In the case represented, the two curves cross twice, thereby implying that there are two BG equilibria featuring positive innovation and growth (points X and Y ). Standard stability analysis is inappropriate for dynamic models with perfect foresight. It is possible, however, to analyze the expectational stability (E-stability) of the BG equilibria. E-stability is tested as follows. Set an arbitrary initial level for the expected interest rate r e , and let agents choose their optimal savings plan according to (83). This implies a notional growth rate of consumption and capital, as determined by the SS curve. Next, firms take action. At the notional growth rate, there is a unique interest
164
G. Gancia and F. Zilibotti
rate consistent with the no-arbitrage condition implied by (82), as shown by the DD curve. The composition of these two operations define a mapping from an expected to a realized interest rate: rt = T rte . (84) A perfect foresight BG equilibrium is a fixed point to the mapping, r = T (r). After consumers have observed the realized interest rate, they update their expectations about next period’s interest rate using adaptive learning, i.e.: e rt+1 (85) = rte + ψt rt − rte , where ξt = ψ/t. The sequence {ψt } determines how sensitive the expectations are to past errors, and it is known as the gain sequence. Substituting (84) into (85) defines a dynamic system, whose stability can be analyzed by linearization techniques. In general, expectational stability occurs whenever T (r) < 1, where r is the steady-state interest rate.60 An inspection of Figure 7 shows the equilibrium X to be E-stable, while the equilibrium Y is not. Let rXe and rYe denote two expected interest rates which are below the equilibria X and Y , respectively. Then, in the case of the equilibrium X, T (rXe ) > rXe , and the adaptive adjustment moves the economy towards the equilibrium, inducing convergence. In contrast, in the case of the equilibrium Y , T (rYe ) < rYe , and the adaptive adjustment moves the economy away from the equilibrium, thereby inducing divergence. In the case analyzed so far, only one BG is E-stable, and E-stability can be used as a selection criteria. It is possible, however, that multiple E-stable BG equilibria exist in the general model with convex adjustment costs. Figure 8 describes a case with four steady-states, two of them being E-stable. Equilibria such as X and Z are E-stable (note that T (rXe ) > rXe and T (rZe ) > rZe ). Moreover, in the neighborhood of these equilibria, there exist stationary sunspot equilibria. In one such equilibrium, the economy switches stochastically between two points in the neighborhood of X and Z, respectively, with switching probabilities given by a time-invariant transition probability matrix. The fact that both X and Z are E-stable is sufficient for any stationary sunspot equilibrium in their neighborhood to be E-stable in itself.61 We conclude that a modified version of the model of growth with expanding variety can generate endogenous fluctuations. The key assumptions are complementarity between capital goods and convex adjustment costs to capital. The former assumption guarantees the existence of multiple BG equilibria, around which sunspot equilibria can be constructed. The latter assumption guarantees that the sunspot equilibrium is expectationally stable, i.e., it can be learned through adaptive expectations. 60 See Evans and Honkapohja (2001) for a state-of-art analysis of expectational indeterminacy. 61 For general discussion of sunspot equilibria, see Azariadis and Guesnerie (1986), Grandmont (1986) and
Azariadis (1993).
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
165
Figure 8.
The model assumes increasing returns to physical and knowledge capital. The reduced form representation of the final good technology is: Y = Aφ−α KFα , where KF = Ax denotes aggregate capital used in intermediate production. Empirical estimates suggest that α 0.4, which implies that the lower bound to the output elasticity of knowledge to generate multiplicity is φ − α = 0.6. Evans et al. (1998) provide a numerical example of an E-stable sunspot equilibrium, assuming φ = 4. Recent estimations from Porter and Stern (2000) using patent numbers report φ − α to be around 0.1, however. Therefore, the model seems to require somehow extreme parameters to generate endogenous fluctuations. Augmenting the model with other accumulated assets, such as human capital, may help obtain the results under realistic parameter configurations. This is complicated by the presence of scale effects in the expanding variety model. However, in a recent paper, Dalgaard and Kreiner (2001) formulate a version of the model with human capital accumulation and without scale effects. In their model, both human capital (embodied knowledge) and technical change (disembodied knowledge) are used to produce final goods. The scale effect is avoided by congestion effects in the accumulation of human capital. An interesting feature of this model is that, unlike other recent models without scale effects, positive long-run growth in income per capita does not hinge on positive population growth.
166
G. Gancia and F. Zilibotti
8. Conclusions In this chapter, we have surveyed recent contributions to growth theory inspired by Romer’s (1990) expanding variety model. Key features of the theory are increasing returns through the introduction of new products that do not displace existing ones and the existence of monopoly rents providing an incentive for firms to undertake costly innovative investments. This model has had a tremendous impact on the literature, and we could only provide a partial review of its applications. Then, we decided to focus on a few major themes: trade and biased technical change, with their effects on growth and inequality, financial development, complementarity in the process of innovation and endogenous fluctuations. While only being a limited selection, these applications give a sense of the success of the model in providing a tractable framework for analyzing a wide array of issues in economic growth. In fact, we have shown how the model can incorporate a number of general equilibrium effects that are fundamental in the analysis of trade, wage inequality, cross-country productivity differences and other topics. Further, while the original model has linear AK-dynamics, we have surveyed recent generalizations featuring richer dynamics, which can potentially be applied to the study of financial development and innovations waves. Given its longevity, flexibility and simplicity, we are convinced that the growth model with horizontal innovation will continue to be useful in future research.
Acknowledgements We thank Philippe Aghion, Jeremy Greenwood, Kiminori Matsuyama and Jerome Vandennbusche for helpful comments, Zheng Song for excellent research assistance, and Christina Lönnblad for editorial assistance.
References Acemoglu, D. (1998). “Why do new technologies complement skills? Directed technical change and wage inequality”. Quarterly Journal of Economics 113, 1055–1090. Acemoglu, D. (2002). “Directed technical change”. Review of Economic Studies 69, 781–809. Acemoglu, D. (2003a). “Patterns of skill premia”. Review of Economic Studies 70, 199–230. Acemoglu, D. (2003b). “Labor- and capital-augmenting technical change”. Journal of the European Economic Association 1, 1–37. Acemoglu, D., Zilibotti, F. (2001). “Productivity differences”. Quarterly Journal of Economics 116, 563–606. Acemoglu, D., Zilibotti, F. (1997). “Was Prometheus unbounded by chance? Risk, diversification and growth”. Journal of Political Economy 105, 710–751. Aghion, P., Howitt, P. (1998). Endogenous Growth Theory. MIT Press, Cambridge, MA. Aghion, P., Howitt, P. (1992). “A model of growth through creative destruction”. Econometrica 60 (2), 323– 351. Antras, P. (2004). “Is the U.S. aggregate production function Cobb–Douglas? New estimates of the elasticity of substitution”. Contributions to Macroeconomics 4 (1).
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
167
Arrow, K.J. (1962). “The economic implications of learning-by-doing”. Review of Economic Studies 29 (1), 155–173. Atkinson, A.B., Stiglitz, J.E. (1969). “A new view of technological change”. Economic Journal, 573–578. Azariadis, C. (1993). Intertemporal Macroeconomics. Blackwell. Azariadis, C., Guesnerie, R. (1986). “Sunspots and cycles”. Review of Economic Studies 53, 725–737. Barro, R.J., Sala-i-Martin, X. (1995). Economic Growth. McGraw-Hill. Barro, R.J., Sala-i-Martin, X. (1997). “Technological diffusion, convergence, and growth”. Journal of Economic Growth 2, 1–26. Benassy, J.-P. (1998). “Is there always too little research in endogenous growth with expanding product variety?”. European Economic Review 42, 61–69. Bencivenga, V., Smith, B. (1991). “Financial intermediation and endogenous growth”. Review of Economic Studies 58, 195–209. Bessen, J., Maskin, E. (2002). “Sequential innovation, patents and imitation”. Mimeo. Princeton. Borjas, G.J., Freeman, R.B., Katz, L.F. (1997). “How much do immigration and trade affect labor market outcomes”. Brookings Papers on Economic Activity, 1–67. Braudel, F. (1979). Civilization and Capitalism. Harper and Row, New York. Caselli, F., Esquivel, G., Lefort, F. (1996). “Reopening the convergence debate: A new look at cross-country growth empirics”. Journal of Economic Growth 1, 363–389. Caselli, F., Coleman, W.J. (2005), “The world technology frontier”. American Economic Review, in press. Cetorelli, N. (2002). “Could Prometheus be bound again? A contribution to the convergence controversy”. Journal of Economic Dynamics and Control 27, 29–50. Ciccone, A., Matsuyama, K. (1999). “Efficiency and equilibrium with dynamic increasing returns due to demand complementarities”. Econometrica 67, 499–525. Ciccone, A., Matsuyama, K. (1996). “Start-up costs and pecuniary externalities as barriers to economic development”. Journal of Development Economics 49, 33–59. Cohen, W.M., Levin, R.C. (1989). “Empirical studies of innovation and market structure”. In: Handbook of Industrial Organization, vol. II. Elsevier, Amsterdam, pp. 1059–1107. Cooper, R.W., John, A. (1988). “Coordinating coordination failures in Keynesian models”. Quarterly Journal of Economics 103, 441–463. Dalgaard, C.-J., Kreiner, C.T. (2001). “Is declining productivity inevitable?”. Journal of Economic Growth 6, 187–203. Dinopoulos, E., Segerstrom, P. (1999). “A Schumpeterian model of protection and relative wages”. American Economic Review 89, 450–472. Dinopoulos, E., Segerstrom, P. (2003). “A theory of North–South trade and globalization”. Mimeo. Stockholm School of Economics. Devereux, M.B., Lapham, B.J. (1994). The stability of economic integration and endogenous growth. Quarterly Journal of Economics 109, 299–305. De Vries, J. (1990). The Economy of Europe in an Age of Crisis; 1600–1750. Cambridge University Press, Cambridge. Dixit, A.K., Stiglitz, J.E. (1977). “Monopolistic competition and optimum product diversity”. American Economic Review 67 (3), 297–308. Diwan, I., Rodrik, D. (1991). “Patents, appropriate technology, and North–South trade”. Journal of International Economics 30, 27–48. Dornbusch, R., Fischer, S., Samuelson, P.A. (1977). “A continuum Ricardian model of comparative advantage, trade and payments”. American Economic Review 67, 823–839. Epifani, P., Gancia, G. (2002). “The skill bias of world trade”. IIES Seminar Paper 707. Ethier, W.J. (1982). “National and international returns to scale in the modern theory of international trade”. American Economic Review 72 (3), 389–405. Evans, G., Honkapohja, S., Romer, P. (1998). “Growth cycles”. American Economic Review 88 (3), 495–515. Evans, G., Honkapohja, S. (2001). Learning and Expectations in Macroeconomics. Princeton University Press, Princeton.
168
G. Gancia and F. Zilibotti
Francois, P., Lloyd-Ellis, H. (2003). “Animal spirits through creative destruction”. American Economic Review 93, 530–550. Funke, M., Strulik, H. (2000). “On endogenous growth with physical capital, human capital and product variety”. European Economic Review 44, 491–515. Gancia, G. (2003). “Globalization, divergence and stagnation”. IIES Seminar Paper 720. Grandmont, J.-M. (1986). “Periodic and aperiodic behavior in discrete one-dimensional dynamical system”. In: Hildenbrand, W., Mas-Colell, A. (Eds.), Contributions to Mathematical Economics: In Honor of Gerard Debreu. Elsevier, Amsterdam. Greenwood, J., Jovanovic, B. (1990). “Financial development, growth and the distribution of income”. Journal of Political Economy 98, 1067–1107. Griliches, Z., Schmookler, J. (1963). “Inventing and maximizing”. American Economic Review 53, 725–729. Grossman, G., Helpman, E. (1989). “Product development and international trade”. Journal of Political Economy 97, 1261–1283. Grossman, G., Helpman, E. (1991a). Innovation and Growth in the World Economy. MIT Press, Cambridge, MA. Grossman, G., Helpman, E. (1991b). “Endogenous product cycles”. Economic Journal 101, 1214–1229. Grossman, G., Lai, E. (2004). “International protection of intellectual property”. American Economic Review 94, 1635–1653. Habakkuk, H.J. (1962). American and British Technology in the Nineteenth Century: Search for Labor Saving Inventions. Cambridge University Press, Cambridge. Hall, R., Jones, C.I. (1999). “Why do some countries produce so much more output per worker than others?”. Quarterly Journal of Economics 114, 83–116. Hamermesh, D.S. (1993). Labor Demand. Princeton University Press, Princeton. Helpman, E. (1993). “Innovation, imitation and intellectual property rights”. Econometrica 61, 1247–1280. Helpman, E. (1998). General Purpose Technology and Economic Growth. MIT Press, Cambridge, MA. Helpman, E., Krugman, P. (1985). Market Structure and Foreign Trade. MIT Press, Cambridge, MA. Hicks, J. (1932). The Theory of Wages. Macmillan, London. Imbs, J., Wacziarg, R. (2003). “Stages of diversification”. American Economic Review 93, 63–86. Jones, C.I. (1995). “R&D-based models of economic growth”. Journal of Political Economy 103 (August), 759–784. Judd, K.L. (1985). “On the performance of patents”. Econometrica 53 (3), 567–585. Kalemli-Ozcan, S., Sorensen, B.E., Yosha, O. (2001). “Economic integration, industrial specialization, and the asymmetry of macroeconomic fluctuations”. Journal of International Economics 55, 107–137. Kalemli-Ozcan, S., Sorensen, B.E., Yosha, O. (2003). “Risk sharing and industrial specialization: Regional and international evidence”. American Economic Review 93, 903–918. Kennedy, C. (1964). “Induced bias in innovation and the theory of distribution”. Economic Journal 74, 541– 547. Klenow, P.J. (1996). “Industry innovation: Where and why”. Carnegie–Rochester Conference Series on Public Policy 44, 125–150. Klenow, P.J., Rodriguez-Clare, A. (1997). “The neoclassical revival in growth economics: Has it gone too far?”. In: Bernanke, B.S., Rotemberg, J.J. (Eds.), NBER Macroeconomics Annual 1997. MIT Press, Cambridge, MA, pp. 73–102. Krugman, P. (1991). “Increasing returns and economic geography”. Journal of Political Economy 99, 483– 499. Krugman, P., Venables, A. (1995). “Globalization and the inequality of nations”. Quarterly Journal of Economics 110, 857–880. Krusell, P., Ohanian, L., Rios-Rull, J.-V., Violante, G. (2000). “Capital skill complementarity and inequality”. Econometrica 68, 1029–1053. Kwan, F.Y.K., Lai, E.L.C. (2003). “Intellectual property rights protection and endogenous economic growth”. Journal of Economic Dynamics and Control 27, 853–873. Lai, E.L.C. (1998). “International intellectual property rights protection and the rate of product innovation”. Journal of Development Economics 55, 133–153.
Ch. 3: Horizontal Innovation in the Theory of Growth and Development
169
Lai, E.L.C., Qiu, L.D. (2003). “The North’s intellectual property rights standard for the South?”. Journal of International Economics 59, 183–209. Levhari, D., Sheshinski, E. (1969). “A theorem on returns to scale and steady-state growth”. Journal of Political Economy 77, 60–65. Martin, P., Rey, H. (2000). “Financial integration and asset returns”. European Economic Review 44, 1327– 1350. Martin, P., Rey, H. (2001). “Financial super-markets: Size matters for asset trade”. NBER Working Paper 8476. Martin, P., Rey, H. (2002). “Financial globalization and emerging markets: With or without crash?”. NBER Working Paper 9288. Matsuyama, K. (1995). “Complementarities and cumulative processes in models of monopolistic competition”. Journal of Economic Literature 33, 701–729. Matsuyama, K. (1996). “Why are there rich and poor countries? Symmetry-breaking in the world economy”. Journal of the Japanese and International Economies 10, 419–439. Matsuyama, K. (1997). “Complementarity, instability and multiplicity”. The Japanese Economic Review 48, 240–266. Matsuyama, K. (1999). “Growing through cycles”. Econometrica 67, 335–348. Matsuyama, K. (2001). “Growing through cycles in an infinitely lived agent economy”. Journal of Economic Theory 100, 220–234. Murphy, K.M., Shleifer, A., Vishny, R.W. (1989). “Industrialization and the big push”. Quarterly Journal of Economics 106 (2), 503–530. Neary, P. (2003). “Globalisation and market structure”. Journal of the European Economic Association 1, 245–271. Nordhaus, W.D. (1969a). “An economic theory of technological change”. American Economic Review 59 (2), 18–28. Nordhaus, W.D. (1969b). Invention, Growth and Welfare: A Theoretical Treatment of Technological Change. MIT Press, Cambridge, MA. North, D., Thomas, R.P. (1973). The Rise of the Western World: A New Economic History. Cambridge University Press, Cambridge. Parente, S.L., Prescott, E.C. (1994). “Barriers to technology adoption and development”. Journal of Political Economy 102, 298–321. Porter, M.E., Stern, S. (2000). “Measuring the “ideas” production function: Evidence from international patent output”. NBER Working Paper 7891. Prescott, E.C. (1998). “Needed: A theory of total factor productivity”. International Economic Review 39, 525–553. Ramey, G., Ramey, V. (1995). “Cross-country evidence of the link between volatility and growth”. American Economic Review 85, 1138–1151. Rebelo, S. (1991). “Long-run policy analysis and long-run growth”. Journal of Political Economy 99 (3), 500–521. Rivera-Batiz, L., Romer, P. (1991a). “Economic integration and endogenous growth”. Quarterly Journal of Economics 106, 531–555. Rivera-Batiz, L., Romer, P. (1991b). “International trade with endogenous technological change”. European Economic Review 35, 971–1004. Rodriguez-Clare, A. (1996). “The division of labor and economic development”. Journal of Development Economics 49, 3–32. Romer, P. (1986). “Increasing returns and long-run growth”. Journal of Political Economy 94, 1002–1037. Romer, P. (1987). “Growth based on increasing returns due to specialization”. American Economic Review 77, 56–62. Romer, P. (1990). “Endogenous technological change”. Journal of Political Economy 98, 71–102. Romer, P. (1994). “New goods, old theory and the welfare costs of trade restrictions”. Journal of Development Economics 43, 5–77.
170
G. Gancia and F. Zilibotti
Rosenberg, N. (1976). Perspectives on Technology. Cambridge University Press, Cambridge. Samuelson, P. (1965). “A theory of induced innovations along Kennedy–Weisacker lines”. Review of Economics and Statistics, 444–464. Schmookler, J. (1966). Invention and Economic Growth. Harvard University Press. Schumpeter, J.A. (1942). Capitalism, Socialism and Democracy. Harper, New York. Scotchmer, S. (1996). “Protecting early innovators: Should second-generation products be patentable?” The Rand Journal of Economics 27, 322–331. Scotchmer, S., Green, J. (1995). “On the division of profits between sequential innovators”. The Rand Journal of Economics 26, 20–33. Segerstrom, P.S., Anant, T.C.A., Dinopoulos, E. (1990). “A Schumpeterian model of the product life cycle”. American Economic Review 80, 1077–1091. Shell, K. (1973). “Inventive activity, industrial organization and economic growth”. In: Mirrels, J.A., Stern, N.H. (Eds.), Models of Economic Growth. Wiley, New York. Shleifer, A. (1986). “Implementation cycles”. Journal of Political Economy 94, 1163–1190. Solow, R.M. (1956). “A contribution to the theory of economic growth”. Quarterly Journal of Economics 70, 65–94. Spence, M. (1976). “Product selection, fixed costs, and monopolistic competition”. Review of Economic Studies 43, 217–235. Stiglitz, J.E. (1970). “Factor price equalization in a dynamic economy”. Journal of Political Economy 78, 456–488. Swan, T.W. (1956). “Economic growth and capital accumulation”. Economic Record 32, 334–361. Thoenig, M., Verdier, T. (2003). “Trade induced technical bias and wage inequalities: A theory of defensive innovations”. American Economic Review 93, 709–728. Ventura, J. (1997). “Growth and interdependence”. Quarterly Journal of Economics 112, 57–84. Vernon, R. (1966). “International investment and international trade in product-cycle”. Quarterly Journal of Economics 80, 190–207. Wood, A. (1994). North–South Trade, Employment and Inequality: Changing Fortunes in a Skill Driven World. Clarendon Press, Oxford. Yang, G., Maskus, K.E. (2001). “Intellectual property rights, licensing and innovation in an endogenous product-cycle model”. Journal of International Economics 53, 169–187. Young, A. (1928). “Increasing returns and economic progress”. Economic Journal 38 (152), 527–542. Young, A. (1991). “Learning by doing and the dynamic effects of international trade”. Quarterly Journal of Economics 106, 369–406. Young, A. (1993). “Substitution and complementarity in endogenous innovation”. Quarterly Journal of Economics 108, 775–807. Young, A. (1995). “The tyranny of numbers: Confronting the statistical realities of the East Asian growth experience”. Quarterly Journal of Economics 110, 641–680. Xie, D. (1998). “An endogenous growth model with expanding ranges of consumer goods and of producer durables”. International Economic Review 39, 439–460. Zilibotti, F. (1994). “Endogenous growth and intermediation in an archipelago economy”. Economic Journal 104, 462–473. Zilibotti, F. (1995). “A Rostovian model of endogenous growth and underdevelopment traps”. European Economic Review 39, 1569–1602.
Chapter 4
FROM STAGNATION TO GROWTH: UNIFIED GROWTH THEORY ODED GALOR Brown University and Hebrew University
Contents Abstract Keywords 1. Introduction 2. Historical evidence 2.1. The Malthusian epoch 2.1.1. Income per capita 2.1.2. Income and population growth 2.2. The Post-Malthusian Regime 2.2.1. Income per capita 2.2.2. Income and population growth 2.2.3. Industrialization and urbanization 2.2.4. Early stages of human capital formation 2.3. The Sustained Growth Regime 2.3.1. Growth of income per capita 2.3.2. The demographic transition 2.3.3. Industrial development and human capital formation 2.3.4. International trade and industrialization 2.4. The great divergence
3. The fundamental challenges 3.1. Mysteries of the growth process 3.2. The incompatibility of non-unified growth theories 3.2.1. Malthusian and Post-Malthusian theories 3.2.2. Theories of modern economic growth 3.3. Theories of the demographic transition and their empirical assessment 3.3.1. The decline in infant and child mortality 3.3.2. The rise in the level of income per capita 3.3.3. The rise in the demand for human capital 3.3.4. The decline in the gender gap 3.3.5. Other theories
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01004-X
172 173 174 178 179 180 180 185 186 188 191 194 195 195 198 204 216 218 219 220 221 221 223 224 225 227 229 232 234
172
O. Galor
4. Unified growth theory 4.1. From stagnation to growth 4.1.1. Central building blocks 4.1.2. The basic structure of the model 4.1.3. The dynamical system 4.1.4. From Malthusian stagnation to sustained growth 4.1.5. Major hypotheses and their empirical assessment 4.2. Complementary theories 4.2.1. Alternative mechanisms for the emergence of human capital formation 4.2.2. Alternative triggers for the demographic transition 4.2.3. Alternative modeling of the transition from agricultural to industrial economy
5. Unified evolutionary growth theory 5.1. Human evolution and economic development 5.2. Natural selection and the origin of economic growth 5.2.1. Primary ingredients 5.2.2. Main hypotheses and their empirical assessment 5.3. Complementary mechanisms 5.3.1. The evolution of ability and economic growth 5.3.2. The evolution of life expectancy and economic growth
6. Differential takeoffs and the great divergence 6.1. Non-unified theories 6.2. Unified theories 6.2.1. Human capital promoting institutions 6.2.2. Globalization and the great divergence
7. Concluding remarks Acknowledgements References
235 237 238 239 245 251 253 256 257 260 262 264 264 266 268 271 273 274 274 276 277 279 280 280 283 285 285
Abstract The transition from stagnation to growth and the associated phenomenon of the great divergence have been the subject of an intensive research in the growth literature in recent years. The discrepancy between the predictions of exogenous and endogenous growth models and the process of development over most of human history, induced growth theorists to advance an alternative theory that would capture in a single unified framework the contemporary era of sustained economic growth, the epoch of Malthusian stagnation that had characterized most of the process of development, and the fundamental driving forces of the recent transition between these distinct regimes. The advancement of unified growth theory was fueled by the conviction that the understanding of the contemporary growth process would be limited and distorted unless growth theory would be based on micro-foundations that would reflect the qualitative aspects of the growth process in its entirety. In particular, the hurdles faced by less
Ch. 4: From Stagnation to Growth: Unified Growth Theory
173
developed economies in reaching a state of sustained economic growth would remain obscured unless the origin of the transition of the currently developed economies into a state of sustained economic growth would be identified, and its implications would be modified to account for the additional economic forces faced by less developed economies in an interdependent world. Unified growth theory suggests that the transition from stagnation to growth is an inevitable outcome of the process of development. The inherent Malthusian interaction between the level of technology and the size and the composition of the population accelerated the pace of technological progress, and ultimately raised the importance of human capital in the production process. The rise in the demand for human capital in the second phase of industrialization, and its impact on the formation of human capital as well as on the onset of the demographic transition, brought about significant technological advancements along with a reduction in fertility rates and population growth, enabling economies to convert a larger share of the fruits of factor accumulation and technological progress into growth of income per capita, and paving the way for the emergence of sustained economic growth. Variations in the timing of the transition from stagnation to growth and thus in economic performance across countries reflect initial differences in geographical factors and historical accidents and their manifestation in variations in institutional, social, cultural, and political factors. In particular, once a technologically driven demand for human capital emerged in the second phase of industrialization, the prevalence of human capital promoting institutions determined the extensiveness of human capital formation, the timing of the demographic transition, and the pace of the transition from stagnation to growth.
Keywords growth, technological progress, demographic transition, human capital, evolution, natural selection, Malthusian stagnation JEL classification: O11, O14, O33, O40, J11, J13
174
O. Galor
1. Introduction This chapter examines the recent advance of a unified growth theory that is designed to capture the complexity of the process of growth and development over the entire course of human history. The evolution of economies during the major portion of human history was marked by Malthusian Stagnation. Technological progress and population growth were miniscule by modern standards and the average growth rate of income per capita in various regions of the world was even slower due to the offsetting effect of population growth on the expansion of resources per capita. In the past two centuries, in contrast, the pace of technological progress increased significantly in association with the process of industrialization. Various regions of the world departed from the Malthusian trap and experienced initially a considerable rise in the growth rates of income per capita and population. Unlike episodes of technological progress in the pre-Industrial Revolution era that failed to generate sustained economic growth, the increasing role of human capital in the production process in the second phase of industrialization ultimately prompted a demographic transition, liberating the gains in productivity from the counterbalancing effects of population growth. The decline in the growth rate of population and the associated enhancement of technological progress and human capital formation paved the way for the emergence of the modern state of sustained economic growth. The transitions from a Malthusian epoch to a state of sustained economic growth and the related phenomenon of the Great Divergence, as depicted in Figure 1, have significantly shaped the contemporary world economy.1 Nevertheless, the distinct qualitative aspects of the growth process during most of human history were virtually ignored in the shaping of growth models, resulting in a growth theory that is consistent with a small fragment of human history. The inconsistency of exogenous and endogenous growth models with some of the most fundamental features of process of development, has led recently to a search for a unified theory that would unveil the underlying micro-foundations of the growth process in its entirety, capturing the epoch of Malthusian Stagnation that characterized most of human history, the contemporary era of modern economic growth, and the underlying driving forces that triggered the recent transition between these regimes and the associated phenomenon of the Great Divergence in income per capita across countries. The preoccupation of growth theory with empirical regularities that have characterized the growth process of developed economies in the past century and of less developed economies in the last few decades, has become harder to justify from a scientific viewpoint in light of the existence of vast evidence about qualitatively different 1 The ratio of GDP per capita between the richest region and the poorest region in the world was only 1.1 : 1 in the year 1000, 2 : 1 in the year 1500, and 3 : 1 in the year 1820. In the course of the ‘Great Divergence’ the ratio of GDP per capita between the richest region and the poorest region has widened considerably from a modest 3 : 1 ratio in 1820, to a 5 : 1 ratio in 1870, a 9 : 1 ratio in 1913, a 15 : 1 ratio in 1950, and a 18 : 1 ratio in 2001.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
175
Figure 1. The evolution of regional income per capita over the years 1–2000. Source: Maddison (2003).2
empirical regularities that characterized the growth process over most of human existence. It has become evident that in the absence of a unified growth theory that is consistent with the entire process of development, the understanding of the contemporary growth process would be limited and distorted. As stated eloquently by Copernicus: “It is as though an artist were to gather the hands, feet, head and other members for his images from diverse models, each part perfectly drawn, but not related to a single body, and since they in no way match each other, the result would be monster rather than man.”3 The evolution of theories in older scientific disciplines suggests that theories that are founded on the basis of a subset of the existing observations and their driving forces, may be attractive in the short run, but non-robust and eventually non-durable in the long run.4 The attempts to develop unified theories in physics have been based on the conviction that all physical phenomena should be explainable by some underlying unity.5 Similarly, the entire process of development and its foundamental forces ought to be captured by a unified growth theory.
2 According to Maddison’s classification, “Western Offshoots” consist of the United States, Canada, Australia and New Zealand. 3 Quoted in Kuhn (1957). 4 For instance, Classical Thermodynamics that lacked micro-foundations was ultimately superseded by the micro-based Statistical Mechanics. 5 Unified Field Theory, for instance, proposes to unify by a set of general laws the four distinct forces that are known to control all the observed interactions in matter: electromagnetism, gravitation, the weak force, and the strong force. The term ‘unified field theory’ was coined by Einstein, whose research on relativity led him to the hypothesis that it should be possible to find a unifying theory for the electromagnetic and gravitational forces.
176
O. Galor
The transition from stagnation to growth and the associated phenomenon of the great divergence have been the subject of intensive research in the growth literature in recent years.6 It has been increasingly recognized that the understanding of the contemporary growth process would be fragile and incomplete unless growth theory could be based on proper micro-foundations that would reflect the various qualitative aspects of the growth process and their central driving forces. Moreover, it has become apparent that a comprehensive understanding of the hurdles faced by less developed economies in reaching a state of sustained economic growth would be futile unless the factors that prompted the transition of the currently developed economies into a state of sustained economic growth could be identified and their implications would be modified to account for the differences in the growth structure of less developed economies in an interdependent world. Imposing the constraint that a single theory should account for the entire intricate process of development and its prime causes in the last thousands of years is a discipline that would enhance the viability of growth theory. A unified theory of economic growth would reveal the fundamental micro-foundations that are consistent with the process of economic development over the entire course of human history, rather that with the last century only, boosting the confidence in growth theory, its predictions and policy implications. Moreover, it would improve the understanding of the underlying factors that led to the transition from stagnation to growth of the currently developed countries, shedding light on the growth process of the less developed economies. The establishment of a unified growth theory has been a great intellectual challenge, requiring major methodological innovations in the construction of dynamical systems that could capture the complexity which characterized the evolution of economies from a Malthusian epoch to a state of sustained economic growth. Historical evidence suggests that the transition from the Malthusian epoch to a state of sustained economic growth, rapid as it may appear, was a gradual process and thus could not plausibly be viewed as the outcome of a major exogenous shock that shifted economies from the basin of attraction of the Malthusian epoch into the basin of attraction of the Modern Growth Regime.7 The simplest methodology for the generation of a phase transition – a major shock in an environment characterized by multiple locally stable equilibria –
6 The transition from Malthusian stagnation to sustained economic growth was explored by Galor and Weil (1999, 2000), Lucas (2002), Galor and Moav (2002), Hansen and Prescott (2002), Jones (2001), as well as others, and the association of Great Divergence with this transition was analyzed by Galor and Mountford (2003). 7 As established in Section 2, and consistently with the revisionist view of the Industrial Revolution, neither the 19th century’s take-off of the currently developed world, nor the recent take-off of less developed economies provide evidence for an unprecedented shock that generated a quantum leap in income per-capita. In particular, technological progress could not be viewed as a shock to the system. As argued by Mokyr (2002) technological progress during the Industrial Revolution was an outcome of a gradual endogenous process that took place over this time period. Similarly, technological progress in less developed economies was an outcome of a deliberate decision by entrepreneurs to adopt existing advanced technologies.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
177
was therefore not applicable for the generation of the observed transition from stagnation to growth. An alternative methodology for the observed phase transition was rather difficult to establish since a unified growth theory in which economies take off gradually but swiftly from an epoch of a stable Malthusian stagnation would necessitate a gradual escape from an absorbing (stable) equilibrium – a contradiction to the essence of a stable equilibrium. Ultimately, however, it has become apparent that the observed rapid, continuous, phase transition would be captured by a single dynamical system, if the set of steady-state equilibria and their stability would be altered qualitatively in the process of development. As proposed in unified growth theory, first advanced by Galor and Weil (2000), during the Malthusian epoch the dynamical system would have to be characterized by a stable Malthusian equilibrium, but eventually, due to the evolution of latent state variables the dynamical system would change qualitatively, the Malthusian equilibrium would vanish endogenously, leaving the arena to the gravitational forces of the emerging Modern-Growth Regime, and permitting the economy to take off and to converge to a modern-growth steady-state equilibrium. The observed role of the demographic transition in the shift from the Post-Malthusian Regime to the Sustained Growth Regime and the associated non-monotonic evolution of the relationship between income per capita and population growth added to the complexity of the desirable dynamical system. In order to capture this additional transition unified growth theory had to generate endogenously, in the midst of the process of industrialization, a reversal in the positive Malthusian effect of income on population, providing the reduction in fertility the observed role in the transition to a state of sustained economic growth. As elaborated in this chapter, unified growth theory explores the fundamental factors that generated the remarkable escape from the Malthusian epoch and their significance for the understanding of the contemporary growth process of developed and less developed economies. It deciphers some of the most fundamental questions that have been shrouded in mystery: what accounts for the epoch of stagnation that characterized most of human history? what is the origin of the sudden spurt in growth rates of output per capita and population? why had episodes of technological progress in the pre-industrialization era failed to generate sustained economic growth? what was the source of the dramatic reversal in the positive relationship between income per capita and population that existed throughout most of human history? what triggered the demographic transition? would the transition to a state of sustained economic growth have been feasible without the demographic transition? and, what are the underlying behavioral and technological structures that could simultaneously account for these distinct phases of development and what are their implications for the contemporary growth process of developed and underdeveloped countries? Moreover, unified growth theory sheds light on the perplexing phenomenon of the Great Divergence in income per capita across regions of the world in the past two centuries: what accounts for the sudden take-off from stagnation to growth in some countries in the world and the persistent stagnation in others? why has the positive
178
O. Galor
link between income per capita and population growth reversed its course in some economies but not in others? why have the differences in income per capita across countries increased so markedly in the last two centuries? and has the transition to a state of sustained economic growth in advanced economies adversely affected the process of development in less-developed economies? Unified growth theory suggests that the transition from stagnation to growth is an inevitable by-product of the process of development. The inherent Malthusian interaction between technology and population, accelerated the pace of technological progress, and ultimately brought about an industrial demand for human capital, stimulating human capital formation, and thus further technological progress, and triggering a demographic transition, that has enabled economies to convert a larger share of the fruits of factor accumulation and technological progress into growth of income per capita. Moreover, the theory suggests that differences in the timing of the take-off from stagnation to growth across countries contributed significantly to the Great Divergence and to the emergence of convergence clubs. Variations in the economic performance across countries and regions (e.g., earlier industrialization in England than in China) reflect initial differences in geographical factors and historical accidents and their manifestation in variations in institutional, demographic, and cultural factors, trade patterns, colonial status, and public policy.
2. Historical evidence This section examines the historical evidence about the evolution of the relationship between income per capita, population growth, technological change and human capital formation during the course of three distinct regimes that have characterized the process of economic development: The Malthusian Epoch, The Post-Malthusian Regime, and the Sustained Growth Regime. During the Malthusian Epoch that characterized most of human history, technological progress and population growth were insignificant by modern standards. The level of income per capita had a positive effect on population and the average growth rate of income per capita in the long-run, as depicted in Figure 2, was negligible due to the slow pace of technological progress as well as the counterbalancing effect of population growth on the expansion of resources per capita. During the Post Malthusian Regime, the pace of technological progress markedly increased in association with the process of industrialization, triggering a take-off from the Malthusian trap. The growth rate of income per capita, as depicted in Figures 1 and 2, increased significantly but the positive Malthusian effect of income per capita on population growth was still maintained, generating a sizable increase in population growth that offset some of the potential gains in income per capita. The acceleration in the rate of technological progress in the second phase of industrialization, and its interaction with human capital formation, eventually prompted the demographic transition. The rise in aggregate income was no longer
Ch. 4: From Stagnation to Growth: Unified Growth Theory
179
Figure 2. The evolution of the world income per capita over the years 1–2000. Sources: Maddison (2001, 2003).
counterbalanced by population growth, enabling technological progress to bring about sustained increase in income per capita. 2.1. The Malthusian epoch During the Malthusian epoch that had characterized most of human history, humans were subjected to persistent struggle for existence. Technological progress was insignificant by modern standards and resources generated by technological progress and land expansion were channeled primarily towards an increase in the size of the population, with a minor long-run effect on income per capita. The positive effect of the standard of living on population growth along with diminishing labor productivity kept income per capita in the proximity of a subsistence level.8 Periods marked by the absence of changes in the level of technology or in the availability of land, were characterized by a stable population size as well as a constant income per capita, whereas periods characterized by improvements in the technological environment or in the availability of land generated temporary gains in income per capita, leading eventually to a larger but not richer population. Technologically superior countries had eventually denser populations but their standard of living did not reflect the degree of their technological advancement.9
8 This subsistence level of consumption may be well above the minimal physiological requirements that are
necessary in order to sustain an active human being. 9 Thus, as reflected in the viewpoint of a prominent observer of the period, “the most decisive mark of the prosperity of any country [was] the increase in the number of its inhabitants” [Smith (1776)].
180
O. Galor
2.1.1. Income per capita During the Malthusian epoch the average growth rate of output per capita was negligible and the standard of living did not differ greatly across countries. As depicted in Figure 2 the average level of income per capita during the first millennium fluctuated around $450 per year, and the average growth rate of output per capita in the world was nearly zero.10 This state of Malthusian Stagnation persisted until the end of the 18th century. In the years 1000–1820, the average level of income per capita in the world economy was below $670 per year and the average growth rate of the world income per capita was miniscule, creeping at a rate of about 0.05% per year [Maddison (2001)]. This pattern of stagnation was observed across all regions of the world. As depicted in Figure 1, the average level of income per capita in Western and Eastern Europe, the Western Offshoots, Asia, Africa, and Latin America was in the range of $400–450 per year in the first millennium and the average growth rate in each of these regions was nearly zero. This state of stagnation persisted until the end of the 18th century across all regions and the level of income per capita in 1820 ranged from $418 per year in Africa, $581 in Asia, $692 in Latin America, and $683 in Eastern Europe, to $1202 in the Western Offshoots (i.e., United States, Canada, Australia and New Zealand), and $1204 in Western Europe. Furthermore, the average growth rate of output per capita over this period ranged from 0% in the impoverished region of Africa to a sluggish rate of 0.14% in the prosperous region of Western Europe. Despite the stability in the evolution of the world income per capita in the Malthusian epoch, from a perspective of a millennium, wages and income per capita fluctuated significantly within regions deviating from their sluggish long-run trend over decades and sometimes over several centuries. In particular, as depicted in Figure 3, real GDP per capita in England fluctuated drastically over most of the past millennium. It declined during the 13th century, and increased sharply during the 14th and 15th centuries in response to the catastrophic population decline in the aftermath of the Black Death. This two-century rise in per capita real income stimulated population growth and brought about a decline in income per capita in the 16th century, back to its level in the first half of the 14th century. Real income per capita increased once again in the 17th century and remained stable during the 18th century, prior to its rise during the take-off in the 19th century. 2.1.2. Income and population growth Population growth and the level of income Population growth over this era followed the Malthusian pattern as well. As depicted in Figures 4 and 5, the slow pace of resource expansion in the first millennium was reflected in a modest increase in the population of the world from 231 million people in 1 AD to 268 million in 1000 AD; a miniscule
10 Maddison’s estimates of income per capita are evaluated in terms of 1990 international dollars.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
181
Figure 3. Fluctuations in real GDP per capita: England, 1260–1870. Source: Clark (2001).
Figure 4. The evolution of world population and income per capita over the years 1–2000. Source: Maddison (2001).
average growth rate of 0.02% per year.11 The more rapid (but still very slow) expansion of resources in the period 1000–1500, permitted the world population to increase by 63% over this period, from 268 million in 1000 AD to 438 million in 1500; a slow 0.1% average growth rate per year. Resource expansion over the period 1500–1820 had a more significant impact on the world population, which grew 138% from 438 million 11 Since output per capita grew at an average rate of 0% per year over the period 0–1000, the pace of resource
expansion was approximately equal to the pace of population growth, namely, 0.02% per year.
182
O. Galor
Figure 5. Population growth and income per capita in the world economy. Source: Maddison (2001).
in 1500 to 1041 million in 1820; an average pace of 0.27% per year.12 This positive effect of income per capita on the size of the population was maintained in the last two centuries as well, as world population reached a remarkable level of nearly 6 billion people. Moreover, the gradual increase in income per capita during the Malthusian epoch was associated with a monotonic increase in the average rate of growth of world population, as depicted in Figure 5. This pattern was exhibited within and across countries.13 Fluctuations in income and population Fluctuations in population and wages over this epoch exhibited the Malthusian pattern as well. Episodes of technological progress, land expansion, favorable climatic conditions, or major epidemics (that resulted in a decline of the adult population), brought about a temporary increase in real wages and income per capita. In particular, as depicted in Figure 6, the catastrophic decline in the population of England during the Black Death (1348–1349), from about 6 million
12 Since output per capita in the world grew at an average rate of 0.05% per year in the time period 1000–
1500 as well as in the period 1500–1820, the pace of resource expansion was approximately equal to the sum of the pace of population growth and the growth of output per capita. Namely, 0.15% per year in the period, 1000–1500 and 0.32% per year in the period 1500–1820. 13 Lee (1997) reports positive income elasticity of fertility and negative income elasticity of mortality from studies examining a wide range of pre-industrial countries. Similarly, Wrigley and Schofield (1981) find a strong positive correlation between real wages and marriage rates in England over the period 1551–1801. Clark and Hamilton (2003) find that in England, at the beginning of the 17th century, the number of surviving offspring is higher among households with higher level of income and literacy rates, suggesting that the positive effect of income on fertility is present cross-sectionally, as well.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
183
Figure 6. Population and real wages: England, 1250–1750. Sources: Clark (2001, 2002).
to about 3.5 million people, increased significantly the land–labor ratio, tripling real wages in the subsequent 150 years. Ultimately, however, most of this increase in real resources per capita was channeled towards increased fertility rates, increasing the size of the population, and bringing the real wage rate in the 1560s back to the proximity of its pre-plague level.14 Population density Variations in population density across countries during the Malthusian epoch reflected primarily cross country differences in technologies and land productivity. Due to the positive adjustment of population to an increase in income per capita, differences in technologies or in land productivity across countries resulted in variations in population density rather than in the standard of living.15 For instance, China’s technological advancement in the period 1500–1820 permitted its share of world population to increase from 23.5% to 36.6%, while its income per capita in the beginning and the end of this time interval remained constant at roughly $600 per year.16 This pattern of increased population density persisted until the demographic transition, namely, as long as the positive relationship between income per capita and 14 Reliable population data is not available for the period 1405–1525 and Figure 6 is depicted under the
assumption maintained by Clark (2001) that population was rather stable over this period. 15 Consistent with the Malthusian paradigm, China’s sophisticated agricultural technologies, for example,
allowed high per-acre yields, but failed to raise the standard of living above subsistence. Similarly, the introduction of the potato in Ireland, in the middle of the 17th century, generated a large increase in population over two centuries without significant improvements in the standard of living. Furthermore, the destruction of potatoes by fungus in the middle of the 19th century, generated a massive decline in population due to the Great Famine and mass migration [Mokyr (1985)]. 16 The Chinese population more than tripled over this period, increasing from 103 million in 1500 to 381 million in 1820.
184
O. Galor
Figure 7. Fertility and mortality: England 1540–1820. Source: Wrigley and Schofield (1981).
population growth was maintained. In the period 1600–1870, United Kingdom’s technological advancement relative to the rest of the world more than doubled its share of world population from 1.1% to 2.5%. Similarly, in the period 1820–1870, the land abundant, technologically advanced economy of the US experienced a 220% increase in its share of world population from 1% to 3.2%.17 Mortality and fertility The Malthusian demographic regime was characterized by fluctuations in fertility rates, reflecting variability in income per capita as well as changes in mortality rates. The relationship between fertility and mortality during the Malthusian epoch was complex. Periods of rising income per capita permitted a rise in the number of surviving offspring, inducing an increase in fertility rates along with a reduction in mortality rates, due to improved nourishment, and health infrastructure. Periods of rising mortality rates (e.g., the black death) induced an increase in fertility rates so as to maintain the number of surviving offspring that can be supported by existing resources. In particular, demographic patterns in England during the 14th and 15th centuries, as depicted in Figure 6, suggest that an (exogenous) increase in mortality rates was indeed associated with a significant rise in fertility rates. However, the period 1540– 1820 in England, vividly demonstrates a negative relationship between mortality rates and fertility rates. As depicted in Figure 7, an increase in mortality rates over the period 1560–1650 was associated with a decline in fertility rates, whereas the rise in income per capita in the time period 1680–1820 was associated with a decline in mortality rates along with increasing fertility rates. 17 The population of the United Kingdom nearly quadrupled over the period 1700–1870, increasing from 8.6
million in 1700 to 31.4 million in 1870. Similarly, the population of the United states increased 40-fold, from 1.0 million in 1700 to 40.2 million in 1870, due to a significant labor migration, as well as high fertility rates.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
185
Figure 8. Life expectancy: England, 1540–1870. Source: Wrigley and Schofield (1981).
Life expectancy Life expectancy at birth fluctuated in the Malthusian epoch, ranging from 24 in Egypt in the time period 33–258 AD, to 42 in England at the end of the 16th century. In the initial process of European urbanization, the percentage of urban population increased six-fold from about 3% in 1520 to nearly 18% in 1750 [De Vries (1984) and Bairoch (1988)]. This rapid increase in population density, without significant changes in health infrastructure, generated a rise in mortality rates and a decline in life expectancy. As depicted in Figures 7 and 8, over the period 1580–1740 mortality rates increased by 50% and life expectancy at birth fell from about 40 to nearly 30 years [Wrigley and Schofield (1981)]. A decline in mortality along with a rise in life expectancy began in the 1740s. Life expectancy at birth rose from about 30 to 40 in England and from 25 to 40 in France over the period 1740–1830 [Livi-Bacci (1997)]. 2.2. The Post-Malthusian Regime During the Post-Malthusian Regime the pace of technological progress markedly increased along the process of industrialization.18 The growth rate of output per capita increased significantly, as depicted in Figures 1, 2 and 3, but the positive Malthusian effect of income per capita on population growth was still maintained, generating a sizable increase in population growth, as depicted in Figures 4 and 5, and offsetting some of the potential gains in income per capita. The take-off of developed regions from the Malthusian Regime was associated with the Industrial Revolution and occurred at the beginning of the 19th century, whereas the take-off of less developed regions occurred towards the beginning of the 20th century and was delayed in some countries well into the 20th century. The Post-Malthusian 18 Ironically, shortly before the publication of Malthus’ influential essay, some regions in the world began to
emerge from the trap that he described.
186
O. Galor
Regime ended with the decline in population growth in Western Europe and the Western Offshoots (i.e. United States, Canada, Australia and New Zealand) towards the end of the 19th century, and in less developed regions in the second half of the 20th century. 2.2.1. Income per capita During the Post-Malthusian Regime the average growth rate of output per capita increased significantly and the standard of living started to differ considerably across countries. As depicted in Figure 2, the average growth rate of output per capita in the world soared from 0.05% per year in the time period 1500–1820 to 0.53% per year in 1820–1870, and 1.3% per year in 1870–1913. The timing of the take-off and its magnitude differed across regions. As depicted in Figure 9, the take-off from the Malthusian Epoch and the transition to the Post-Malthusian Regime occurred in Western Europe, the Western Offshoots, and Eastern Europe at the beginning of the 19th century, whereas in Latin America, Asia (excluding China) and Africa it took place at the end of the 19th century. Among the regions that took off at the beginning of the 19th century, the growth rate of income per capita in Western Europe increased from 0.15% per year in the years 1500–1820 to 0.95% per year in the time period 1820–1870, and the growth rates of income per capita of the Western Offshoots increased over the corresponding time periods from 0.34% per year to 1.42% per year. In contrast, the take-off in Eastern Europe was more modest, and its growth rate increased from 0.1% per year in the period 1500– 1820 to 0.63% per year in the time interval 1820–1870. Among the regions that took off towards the end of the 19th century, the average growth rate of income per capita in Latin America jumped from a sluggish rate of 0.11% per year in the years 1820–1870 to a considerable rate of 1.81% per year in the time period 1870–1913, whereas Africa’s growth rates increased more modestly from 0.12% per year in the years 1820–1870 to 0.64% per year in time interval 1870–1913 and 1.02% per year in the period 1913–1950. Asia’s (excluding Japan, China and India) take-off was modest as well, increasing from 0.13% per year in the years 1820–1870 to 0.64% per year in the 1870–1913 period.19 The level of income per capita in the various regions of the world, as depicted in Figure 1, ranged in 1870 from $444 in Africa, $543 in Asia, $698 in Latin America, and $871 in Eastern Europe, to $1974 in Western Europe and $2431 in the Western Offshoots. Thus, the differential timing of the take-off from the Malthusian epoch, increased the gap between the richest region of Western Europe and the Western Offshoots to the impoverished region of Africa from about 3 : 1 in 1820 to approximately 5 : 1 in 1870.
19 Japan’s average growth rate increased from 0.19% per year in the period 1820–1870, to 1.48% per year
in the period 1870–1913. India’s growth rate increased from 0% per year to 0.54% per year over the corresponding periods, whereas China’s take-off was delayed till the 1950s.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
187
Figure 9. The differential timing of the take-off across regions. Source: Maddison (2001).
The acceleration in technological progress and the accumulation of physical capital and to a lesser extent human capital, generated a gradual rise in real wages in the urban sector and (partly due to labor mobility) in the rural sector as well. As depicted in Figure 10, the take-off from the Malthusian epoch in the aftermath of the Industrial Revolution was associated in England with a modest rise in real wages in the first decades of the 19th century and a very significant rise in real wages after 1870.20 A very significant rise in real wages was experienced in France, as well, after 1860.
20 Stokey (2001) attributes about half of the rise in real wage over the period 1780–1850 to the forces of
international trade. Moreover, the study suggests that technological change in manufacturing was three times as important as technological change in the energy sector, in contributing to output growth.
188
O. Galor
Figure 10. Real wages in England and France during the take-off from the Malthusian Epoch. Sources: Clark (2002) for England, and Levy-Leboyer and Bourguignon (1990) for France.
2.2.2. Income and population growth The rapid increase in income per capita in the Post-Malthusian Regime was channeled partly towards an increase in the size of the population. During this regime, the Malthusian mechanism linking higher income to higher population growth continued to function, but the effect of higher population on diluting resources per capita, and thus lowering income per capita, was counteracted by the acceleration in technological progress and capital accumulation, allowing income per capita to rise despite the offsetting effects of population growth. The Western European take-off along with that of the Western Offshoots brought about a sharp increase in population growth in these regions and consequently a modest rise in population growth in the world as a whole. The subsequent take-off of less developed regions and the associated increase in their population growth brought about a significant rise in population growth in the world. The rate of population growth in the world increased from an average rate of 0.27% per year in the period 1500–1820 to 0.4% per year in the years 1820–1870, and to 0.8% per year in the time interval 1870–1913. Furthermore, despite the decline in population growth in Western Europe and the Western Offshoots towards the end of the 19th century and the beginning of the 20th century, the delayed take-off of less developed regions and the significant increase in their income per capita prior to their demographic transitions generated a further increase in the rate of population growth in the world to 0.93% per year in the period 1913–1950, and a sharp rise to a high rate of 1.92% per year in the period 1950–1973. Ultimately, the onset of the demographic transition in less developed economies in the second half of the 20th century, gradually reduced population growth rates to 1.66% per year in the 1973–1998 period [Maddison (2001)].
Ch. 4: From Stagnation to Growth: Unified Growth Theory
189
Figure 11. Regional growth of GDP per capita and population: 1500–2000. Source: Maddison (2001).
Growth in income per capita and population growth As depicted in Figure 11, the take-off in the growth rate of income per capita in all regions of the world was associated with a take-off in population growth. In particular, the average growth rates of income per capita in Western Europe over the time period 1820–1870 rose to an annual rate of 0.95% (from 0.15% in the period 1500–1820) along with a significant increase in population growth to an annual rate of about 0.7% (from 0.26% in the period 1500– 1820). Similarly, the average growth rates of income per capita in the Western Offshoots over the years 1820–1870 rose to an annual rate of 1.42% (from 0.34% in the period 1500–1820) along with a significant increase in population growth to an annual rate of about 2.87% (from 0.43% in the period 1500–1820). A similar pattern was observed in Asia, and as depicted in Figure 11, in Africa and Latin America as well. The average growth rates of income per capita in Latin America over the years 1870–1913 rose to an annual rate of 1.81% (from 0.1% in the period 1820–1870) and subsequently to an annual rate of 1.43% in time interval 1913–1950 and 2.52% in the time period 1950–1973 along with a significant increase in popula-
190
O. Galor
tion growth to an annual rate of 1.64% in the period 1870–1913, 1.97% in the years 1913–1950, and 2.73% in the period 1950–1973, prior to the decline in the context of the demographic transition. Similarly, the average growth rates of income per capita in Africa over the 1870–1913 period rose to an annual rate of 0.64% (from 0.12% in the period 1820–1870) and subsequently to an annual rate of 1.02% in the years 1913–1950 and 2.07% in the period 1950–1973, along with a monotonic increase in population growth from a modest average annual rate of 0.4% in the years 1820–1870, to a 0.75% in the years 1870–1913, 1.65% in the years 1913–1950, 2.33% in the time interval 1950–1973, and a rapid average annual rate of 2.73% in the period 1973–1998. Technological leaders and land-abundant regions during the Post-Malthusian era improved their relative position in the world in terms of their level of income per capita as well as their population size. The increase in population density of technological leaders persisted as long as the positive relationship between income per capita and population growth was maintained. Western Europe’s technological advancement relative to the rest of the world increased its share of world population by 16% from 12.8% in 1820 to 14.8% in 1870, whereas the regional technological leader, the United Kingdom, increased its share of world population by 25% (from 2% to 2.5%) over this fifty-year period. Moreover, land abundance and technological advancement in the Western Offshoots (US, Australia, New Zealand and Canada) increased their share of world population by 227% over a fifty-year period, from 1.1% in 1820 to 3.6% in 1870. The rate of population growth relative to the growth rate of aggregate income declined gradually over the period. For instance, the growth rate of total output in Western Europe was 0.3% per year between 1500 and 1700, and 0.6% per year between 1700 and 1820. In both periods, two thirds of the increase in total output was matched by increased population growth, and the growth of income per capita was only 0.1% per year in the earlier period and 0.2% in the later one. In the United Kingdom, where growth was the fastest, the same rough division between total output growth and population growth can be observed: total output grew at an annual rate of 1.1% in the 120 years after 1700, while population grew at an annual rate of 0.7%. Population and income per capita continued to grow after 1820, but increasingly the growth of total output was expressed as growth of income per capita. Population growth was 40% as large as total output growth over the time period 1820–1870, dropping further after the demographic transition to about 20% of output growth over the 1929–1990 period. Fertility and mortality The relaxation in the households’ budget constraint in the PostMalthusian Regime permitted an increase in fertility rates along with an increase in literacy rates and years of schooling. Despite the decline in mortality rates, fertility rates (as well as population growth) increased in most of Western Europe until the second half of the 19th century [Coale and Treadway (1986)].21 In particular, as depicted in Figure 12, in spite of a century of decline in mortality rates, the crude birth rates in
21 See Dyson and Murphy (1985) as well.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
191
Figure 12. Fertility, mortality and net reproduction rate: England, 1730–1871. Source: Wrigley and Schofield (1981).
England increased over the 18th century and the beginning of the 19th century. Thus, the Net Reproduction Rate (i.e., the number of daughters per woman who reach the reproduction age) increased from about the replacement level of 1 surviving daughter per woman in 1740 to about 1.5 surviving daughters per woman in the eve of the demographic transition in 1870. It appears that the significant rise in income per capita in the Post-Malthusian Regime increased the desirable number of surviving offspring and thus, despite the decline in mortality rates, fertility increased significantly so as to enable households to reach this higher desirable level of surviving offspring. Fertility rates and marriage age Fertility was controlled during this period, despite the absence of modern contraceptive methods, partly via adjustment in marriage rates.22 As depicted in Figure 13, increased fertility was achieved by earlier female’s age of marriage, and a decline in fertility by a delay in the marriage age.23 2.2.3. Industrialization and urbanization The take-off of developed and less developed regions from the Malthusian epoch was associated with the acceleration in the process of industrialization as well as with a significant rise in urbanization. 22 The importance of this mechanism of fertility control is reflected in the assertion by William Cobbett
(1763–1835) – a leader of the campaign against the changes brought by the Industrial Revolution – “. . . men, who are able and willing to work, cannot support their families, and ought . . . to be compelled to lead a life of celibacy, for fear of having children to be starved”. 23 The same pattern is observed in the relationship between Crude Birth Rates and Crude Marriage Rates (per 1000).
192
O. Galor
Figure 13. Fertility rates and female’s age of marriage. Source: Wrigley and Schofield (1981).
Figure 14. Per capita levels of industrialization: (UK in 1900 = 100). Source: Bairoch (1982).24
Industrialization The take-off in the developed regions was accompanied by a rapid process of industrialization. As depicted in Figure 14, per-capita level of industrialization (measuring per capita volume of industrial production) increased significantly in the United Kingdom since 1750, rising 50% over the 1750–1800 period, quadrupling in the years 1800–1860, and nearly doubling in the time period 1860–1913. Similarly per-capita level of industrialization accelerated in the United States, doubling in the
24 Notes: Countries are defined according to their 1913 boundaries. Germany from 1953 is defined as East
and West Germany. India after 1928 includes Pakistan.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
193
Figure 15. Percentage of the population in cities with population larger than 10,000. Europe: 1490–1910. Sources: Bairoch (1988) and De Vries (1984).
1750–1800 as well as 1800–1860 periods, and increasing six-fold in the years 1860– 1913. A similar pattern was experienced in Germany, France, Sweden, Switzerland, Belgium, and Canada. Industrialization nearly doubled in the 1800–1860 period, further accelerating in the time interval 1860–1913. The take-off of less developed economies in the 20th century was associated with increased industrialization as well. However, as depicted in Figure 14, during the 19th century these economies experienced a decline in per capita industrialization (i.e., per capita volume of industrial production), reflecting the adverse effect of the sizable increase in population on the level of industrial production per capita (even in the absence of an absolute decline in industrial production) as well as the forces of globalization and colonialism, that induced less developed economies to specialize in the production of raw materials.25 Urbanization The take-off from Malthusian stagnation and the acceleration in the process of industrialization increased significantly the process of urbanization. As reflected in Figure 15, the percentage of the population that lived in European cities with a population larger than 10,000 people nearly tripled over the years 1750–1870, from 17% to 54%. Similarly, the percentage of the population in England that lived in cities with population larger than 5,000 quadrupled over the 1750–1910 period, from 18% to 75% [Bairoch (1988)]. 25 The sources of the decline in the industrialization of less developed economies are explored by Galor and
Mountford (2003). The effect of colonialism on the patterns of production and thus trade is examined by Acemoglu, Johnson and Robinson (2002) and Bertocchi and Canova (2002).
194
O. Galor
Figure 16. The decline in the percentage of agricultural production in total output: Europe: 1790–1910. Source: Mitchell (1981).
This rapid processes of industrialization and urbanization was accompanied by a rapid decline in the share of agricultural production in total output, as depicted in Figure 16. For instance, this share declined in England from 40% in 1790 to 7% in 1910. 2.2.4. Early stages of human capital formation The acceleration in technological progress during the Post-Malthusian Regime and the associated increase in income per capita stimulated the accumulation of human capital in the form of literacy rates, schooling, and health. The increase in the investment in human capital was induced by the gradual relaxation in households’ budget constraints (as reflected by the rise in real wages and income per capita), as well as by qualitative changes in the economic environment that increased the demand for human capital and induced households to invest in the education of their offspring. In the first phase of the Industrial Revolution, human capital had a limited role in the production process. Education was motivated by a variety of reasons, such as religion, enlightenment, social control, moral conformity, sociopolitical stability, social and national cohesion, and military efficiency. The extensiveness of public education was therefore not necessarily correlated with industrial development and it differed across countries due to political, cultural, social, historical and institutional factors. In the second phase of the Industrial Revolution, however, the demand for education increased, reflecting the increasing skill requirements in the process of industrialization.26 26 Evidence suggests that in Western Europe, the economic interests of capitalists were a significant driving
force behind the implementation of educational reforms, reflecting the interest of capitalists in human capital formation and thus in the provision of public education [Galor and Moav (2006)].
Ch. 4: From Stagnation to Growth: Unified Growth Theory
195
During the Post-Malthusian Regime, the average number of years of schooling in England and Wales rose from 2.3 for the cohort born between 1801 and 1805, to 5.2 for the cohort born in the years 1852–1856 [Matthews et al. (1982)]. Furthermore, human capital as reflected by the level of health of the labor force increased over this period. In particular, between 1740 and 1840 life expectancy at birth rose from 33 to 40 in England (Figure 8), and from 25 to 40 in France. The process of industrialization was eventually characterized by a gradual increase in the relative importance of human capital in less developed economies as well. As documented by Barro and Lee (2000) educational attainment increased significantly across all less developed regions in the Post-Malthusian Regime (that ended with the decline in population growth in the 1970s in Latin America and Asia, and was still in motion in Africa at the end of the 20th century). In particular, the average years of schooling increased from 3.5 in 1960 to 4.4 in 1975 in Latin America, from 1.6 in 1960 to 3.4 in 2000 in Sub-Saharan Africa, and from 1.4 in 1960 to 1.9 in 1975 in South Asia. 2.3. The Sustained Growth Regime The acceleration in technological progress and industrialization in the Post-Malthusian Regime and its interaction with the accumulation of human capital brought about a demographic transition, paving the way to a transition to an era of sustained economic growth. In the post demographic-transition period, the rise in aggregate income due to technological progress and factor accumulation has no longer been counterbalanced by population growth, permitting sustained growth in income per capita in regions that have experienced sustained technological progress and factor accumulation. The transition of the developed regions of Western Europe and the Western Offshoots to the state of sustained economic growth occurred towards the end of the 19th century, whereas the transition of some less developed countries in Asia and Latin America occurred towards the end of the 20th century. Africa, in contrast, is still struggling to make this transition. 2.3.1. Growth of income per capita During the Sustained Growth Regime the average growth rate of output per capita increased significantly along with the decline in population growth. The acceleration in technological progress and the associated rise in the demand for human capital brought about a demographic transition in Western Europe, Western Offshoots, and in many of the less advanced economies, permitting sustained increase in income per capita. Income per capita in the last century has advanced at a stable rate of about 2% per year in Western Europe and the Western Offshoots, as depicted in Figure 17. In contrast, some less developed regions experienced sustained growth rates of output per capita only in the last decades. As depicted in Figure 18, the growth rate of output per capita
196
O. Galor
Figure 17. Sustained economic growth: Western Europe and the Western Offshoots, 1870–2001. Source: Maddison (2003).
Figure 18. Income per capita in Africa, Asia and Latin America, 1950–2001. Source: Maddison (2003).
in Asia has been stable in the last 50 years, the growth rate in Latin America has been declining over this period, and the growth of Africa vanished in the last few decades.27 The transition to a state of sustained economic growth in developed as well as less developed regions was accompanied by a rapid process of industrialization. As depicted in Figure 14, the per capita level of industrialization (measuring per capita volume of industrial production) doubled in the time period 1860–1913 and tripled in the course
27 Extensive evidence about the growth process in the last four decades is surveyed by Barro and Sala-i-
Martin (2003).
Ch. 4: From Stagnation to Growth: Unified Growth Theory
197
Figure 19. The sharp rise in real GDP per capita in the transition to sustained economic growth: England 1435–1915. Sources: Clark (2001) and Feinstein (1972).
of the 20th century. Similarly, the per capita level of industrialization in the United States, increased six-fold over the years 1860–1913, and tripled along the 20th century. A similar pattern was experienced in Germany, France, Sweden, Switzerland, Belgium, and Canada where industrialization increased significantly in the time interval 1860– 1913 as well as over the rest of the 20th century. Moreover, less developed economies that made the transition to a state of sustained economic growth in recent decades have experienced a significant increase in industrialization. The transition to a state of sustained economic growth was characterized by a gradual increase in the importance of the accumulation of human capital relative to physical capital as well as with a sharp decline in fertility rates. In the first phase of the Industrial Revolution (1760–1830), capital accumulation as a fraction of GDP increased significantly whereas literacy rates remained largely unchanged. Skills and literacy requirements were minimal, the state devoted virtually no resources to raise the level of literacy of the masses, and workers developed skills primarily through on-the-job training [Green (1990) and Mokyr (1990, 1993)]. Consequently, literacy rates did not increase during the period 1750–1830 [Sanderson (1995)]. In the second phase of the Industrial Revolution, however, the pace of capital accumulation subsided, the education of the labor force markedly increased and skills became necessary for production. The investment ratio which increased from 6% in 1760 to 11.7% in 1831, remained at around 11% on average in the years 1856–1913 [Crafts (1985) and Matthews et al. (1982)]. In contrast, the average years of schooling of male in the labor force, that did not change significantly until the 1830s, tripled by the beginning of the 20th century [Matthews et al. (1982, p. 573)]. The significant rise in the level of income per capita in England as of 1865, as depicted in Figure 19, was associated with an increase in the standard of living [Voth (2004)], and an increase in school enrollment of 10-year olds from 40% in 1870 to 100% in 1900. Moreover, Total
198
O. Galor
Fertility Rates in England sharply declined over this period from about 5 in 1875, to nearly 2 in 1925. The transition to a state of sustained economic growth in the US, as well, was characterized by a gradual increase in the importance of the accumulation of human capital relative to physical capital. Over the time period 1890–1999, the contribution of human capital accumulation to the growth process in the US nearly doubled whereas the contribution of physical capital declined significantly. Goldin and Katz (2001) show that the rate of growth of educational productivity was 0.29% per year over the 1890–1915 period, accounting for about 11% of the annual growth rate of output per capita over this period.28 In the period 1915–1999, the rate of growth of educational productivity was 0.53% per year accounting for about 20% of the annual growth rate of output per capita over this period. Abramovitz and David (2000) report that the fraction of the growth rate of output per capita that is directly attributed to physical capital accumulation declined from an average of 56% in the years 1800–1890 to 31% in the period 1890–1927 and 21% in the time interval 1929–1966. 2.3.2. The demographic transition The demographic transition swept the world in the course of the last century. The unprecedented increase in population growth during the Post-Malthusian Regime was ultimately reversed and the demographic transition brought about a significant reduction in fertility rates and population growth in various regions of the world, enabling economies to convert a larger share of the fruits of factor accumulation and technological progress into growth of income per capita. The demographic transition enhanced the growth process via three channels: (a) the reduction of the dilution of the stock of capital and land; (b) the enhancement of investment in human capital; (c) the alteration of the age distribution of the population, temporarily increasing the size of the labor force relative to the population as a whole. The decline in population growth The timing of the demographic transition differed significantly across regions. As depicted in Figure 20, the reduction in population growth occurred in Western Europe, the Western Offshoots, and Eastern Europe towards the end of the 19th century and in the beginning of the 20th century, whereas Latin America and Asia experienced a decline in the rate of population growth only in the last decades of the 20th century. Africa’s population growth, in contrast, has been rising steadily, although this pattern is likely to reverse in the near future due to the decline in fertility rates in this region since the 1980s. The Western Offshoots experienced the earliest decline in population growth, from an average annual rate of 2.87% in the period 1820–1870 to an annual average rate of
28 They measure educational productivity by the contribution of education to the educational wage differen-
tials.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
199
Figure 20. The differential timing of the demographic transition across regions. Source: Maddison (2001).
2.07% in the time interval 1870–1913 and 1.25% in the years 1913–1950.29 In Western Europe population growth declined from a significantly lower average level of 0.77% per year in the period 1870–1913 to an average rate of 0.42% per year in the period 1913–1950. A similar reduction occurred in Eastern Europe as well.30 In contrast, in Latin America and Asia the reduction in population growth started to take place in the 1970s, whereas the average population growth in Africa has been rising, despite a modest decline in fertility rates.31 Latin America experienced a decline in population growth from an average annual rate of 2.73% in the years 1950–1973 to an annual average rate of 2.01% in the period 1973–1998. Similarly, Asia (excluding Japan) experienced a decline in population growth from an average annual rate of 2.21% in the time period 1950–1973 to an average annual rate of 1.86% in the 1973–1998 period. The decline in fertility in these less developed regions, however, has been more significant, indicating a sharp forthcoming decline in population growth during the next decades. Africa’s increased resources in the Post-Malthusian Regime, however, have been channeled primarily towards population growth. Africa’s population growth rate has increased monotonically from a modest average annual rate of 0.4% over the years 1820–1870, to a 0.75% in the time interval 1870–1913, 1.65% in the period 1913– 1950, 2.33% in 1950–1973, and a rapid average annual rate of 2.73% in the 1973–1998 period. Consequently, the share of the African population in the world increased by 41%
29 Migration played a significant role in the rate of population growth of these land-abundant countries. 30 A sharper reduction in population growth occurred in the United Kingdom, from 0.87% per year in the
period 1870–1913 to 0.27% per year in the period 1913–1950. 31 As depicted in Figure 21, the decline in Total Fertility Rates in these countries started earlier. The delay in
the decline in population growth could be attributed to an increase in life expectancy as well as an increase in the relative size of cohorts of women in a reproduction age.
200
O. Galor
Figure 21. The evolution of Total Fertility Rate across regions, 1960–1999. Source: World Development Indicators (2001).
in the 60-year period 1913–1973 (from 7% in 1913 to 9.9% in 1973), and an additional 30% in the last 25 years, from 9.9% in 1973 to 12.9% in 1998. Fertility decline The decline in population growth followed the decline in fertility rates. As depicted in Figure 21, Total Fertility Rate over the period 1960–1999 plummeted from 6 to 2.7 in Latin America and declined sharply from 6.14 to 3.14 in Asia.32 Furthermore, Total Fertility Rate in Western Europe and the Western Offshoots declined over this period below the replacement level: from 2.8 in 1960 to 1.5 in 1999 in Western Europe and from 3.84 in 1960 to 1.83 in 1999 in the Western Offshoots [World Development Indicators (2001)]. Even in Africa Total Fertility Rate declined moderately from 6.55 in 1960 to 5.0 in 1999. The demographic transition in Western Europe occurred towards the turn of the 19th century. A sharp reduction in fertility took place simultaneously in several countries in the 1870s, and resulted in a decline of about 1/3 in fertility rates in various states within a 50-year period.33 As depicted in Figure 22, Crude Birth Rates in England declined by 44%, from 36 (per 1000) in 1875, to 20 (per 1000) in 1920. Similarly, live births per 1000 women
32 For a comprehensive discussion of the virtues and drawbacks of the various measures of fertility: TFR,
NNR, and CBR, see Weil (2004). 33 Coale and Treadway (1986) find that a 10% decline in fertility rates was completed in 59% of all European
countries in the time period 1890–1920. In particular, a 10% decline was completed in Belgium in 1881, Switzerland in 1887, Germany in 1888, England and Wales in 1892, Scotland in 1894, Netherlands in 1897, Denmark in 1898, Sweden in 1902, Norway in 1903, Austria in 1907, Hungary in 1910, Finland in 1912, Greece and Italy in 1913, Portugal in 1916, Spain in 1920, and Ireland in 1922.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
201
Figure 22. The demographic transition in Western Europe: Crude Birth Rates and Net Reproduction Rates. Sources: Andorka (1978) and Kuzynski (1969).
aged 15–44 fell from 153.6 in 1871–1880 to 109.0 in 1901–1910 [Wrigley (1969)]. In Germany, Crude Birth Rates declined 37%, from 41 (per 1000) in 1875 to 26 (per 1000) in 1920. Sweden’s Crude Birth Rates declined 32%, from 31 (per 1000) in 1875 to 21 (per 1000) in 1920, and in Finland they declined 32%, from 37 (per 1000) in 1875 to 25 (per 1000) in 1920. Finally, although the timing of demographic transition in France represents an anomaly, starting in the second half of the 18th century, France experienced an additional significant reduction in fertility in the time period 1865–1910, and its Crude Birth Rates declined by 26%, from 27 (per 1000) in 1965 to 20 (per 1000) in 1910. The decline in the crude birth rates in the course of the demographic transition was accompanied by a significant decline in the Net Reproduction Rate (i.e., the number of daughters per woman who reach the reproduction age), as depicted in Figure 22. Namely, the decline in fertility during the demographic transition outpaced the decline in mortality rates, and brought about a decline in the number of children who survived to their reproduction age. Similar patterns are observed in the evolution of Total Fertility Rates in Western Europe, as depicted in Figure 23. Total Fertility Rates (TFR) peaked in the 1870s and then declined sharply and simultaneously across Western European States. In England, TFR declined by 51%, from 4.94 children in 1875, to 2.4 in 1920. In Germany, TFR declined 57%, from 5.29 in 1885 to 2.26 in 1920. Sweden’s TFR declined 61%, from 4.51 in 1876 to 1.77 in 1931, in Finland they declined 52%, from 4.96 in 1876 to 2.4 in 1931 and in France, where a major decline occurred in the years 1750–1850, an additional decline took place in the same time period from 3.45 in 1880 to 1.65 in 1920.
202
O. Galor
Figure 23. The demographic transition in Western Europe: Total Fertility Rates. Source: Chesnais (1992).
Figure 24. The mortality decline in Western Europe, 1730–1920. Source: Andorka (1978).
Mortality decline The mortality decline preceded the decline in fertility rates in most countries in the world, with the notable exceptions of France and the United States. The decline in mortality rates preceded the decline in fertility rates in Western European countries in the 1730–1920 period, as depicted in Figures 22 and 24. The decline in mortality rates began in England 140 years prior to the decline in fertility, and in Sweden and Finland nearly 100 years prior to the decline in fertility.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
203
Figure 25. The decline in infant mortality rates across regions, 1960–1999. Source: World Development Indicators (2001).
A similar sequence of events emerges from the pattern of mortality and fertility decline in less developed regions. As depicted in Figures 21 and 25, a sharp decline in infant mortality rates as of 1960 preceded the decline in fertility rates in Africa that took place since the 1980s. Moreover, existing evidence indicates a simultaneous reduction in mortality and fertility in the 1960–2000 period in all other regions. Life expectancy The decline in mortality rates in developed countries since the 18th century, as depicted in Figure 24, corresponded to a gradual increase in life expectancy, generating a further inducement for investment in human capital. As depicted in Figure 26, life expectancy at birth in England increased at a stable pace from 32 years in the 1720s to about 41 years in the 1870s. This pace of the rise in life expectancy increased towards the end of the 19th century and life expectancy reached the levels of 50 years in 1906, 60 years in 1930 and 77 years in 1996. Similarly, the significant decline in mortality rates across less developed regions in the past century, corresponded to an increase in life expectancy. As depicted in Figure 27, life expectancy increased significantly in developed regions in the 19th century, whereas the rise in life expectancy in less developed regions occurred throughout the 20th century, stimulating further human capital formation. In particular, life expectancy nearly tripled in the course of the 20th century in Asia, rising from a level of 24 years in 1900 to 66 years in 1999, reflecting the rise in income per capita as well as the diffusion of medical technology. Similarly, life expectancy in Africa more than doubled from 24 years in 1900 to 52 years in 1999. In contrast, the more rapid advancement in income per capita in Latin America generated an earlier rise in longevity. Life expectancy increased modestly during the 19th century and more
204
O. Galor
Figure 26. The evolution of life expectancy: England 1580–1996. Sources: Wrigley and Schofield (1981) for 1726–1871 and Human Mortality Database (2003) for 1876–1996.
Figure 27. The evolution of life expectancy across regions, 1820–1999. Source: Maddison (2001).
significantly in the course of the 20th century, from 35 years in 1900 to 69 years in 1999. 2.3.3. Industrial development and human capital formation The process of industrialization was characterized by a gradual increase in the relative importance of human capital in the production process. The acceleration in the rate
Ch. 4: From Stagnation to Growth: Unified Growth Theory
205
of technological progress increased gradually the demand for human capital, inducing individuals to invest in education, and stimulating further technological advancement. Moreover, in developed as well as less developed regions, the onset of the process of human capital accumulation preceded the onset of the demographic transition, suggesting that the rise in the demand for human capital in the process of industrialization and the subsequent accumulation of human capital played a significant role in the demographic transition and the transition to a state of sustained economic growth. Developed economies In the first phase of the Industrial Revolution, the extensiveness of the provision of public education was not correlated with industrial development and it differed across countries due to political, cultural, social, historical and institutional factors. Human capital had a limited role in the production process and education served religious, social, and national goals. In contrast, in the second phase of the Industrial Revolution the demand for skilled labor in the growing industrial sector markedly increased. Human capital formation was designed primarily to satisfy the increasing skill requirements in the process of industrialization, and industrialists became involved in shaping the educational system. Notably, the reversal of the Malthusian relation between income and population growth during the demographic transition, corresponded to an increase in the level of resources invested in each child. For example, the literacy rate among men, which was stable at around 65% in the first phase of the Industrial Revolution, increased significantly during the second phase, reaching nearly 100% at the end of the 19th century [Clark (2003)]. In addition, the proportion of children aged 5 to 14 in primary schools increased from 11% in 1855 to 74% in 1900. A similar pattern is observed in other European societies [Flora et al. (1983)]. In particular, as depicted in Figure 28, the proportion of children aged 5 to 14 in primary schools in France increased significantly in the second phase of industrialization, rising from 30% in 1832 to 86% in 1901. Evidence about the evolution of the return to human capital over this period are scarce and controversial. They do not indicate that the skill premium increased markedly in Europe over the course of the 19th century [Clark (2003)]. One can argue that the lack of clear evidence about the increase in the return to human capital over this period is an indication for the absence of a significant increase in the demand for human capital. This partial equilibrium argument, however, is flawed. The return to human capital is affected by the demand and the supply of human capital. Technological progress in the second phase of the Industrial Revolution brought about an increase in the demand for human capital, and indeed, in the absence of a supply response, one would have expected an increase in the return to human capital. However, the significant increase in schooling that took place in the 19th century, and in particular the introduction of public education that lowered the cost of education, generated a significant increase in the supply of educated workers. Some of this supply response was a direct reaction of the increase in the demand for human capital, and thus may only operate to partially offset the increase in the return to human capital. However, the removal of the adverse effect of credit constraints on the acquisition of human capital (as reflected by the introduction
206
O. Galor
Figure 28. The fraction of children age 5–14 in public primary schools, 1820–1940. Source: Flora et al. (1983).
of public education) generated an additional force that increased the supply of educated labor and operated towards a reduction in the return to human capital.34 A. The industrial base for education reforms in the 19th century Education reforms in developed countries in the 18th and 19th centuries provide a profound insight about the significance of industrial development in the formation of human capital (and thus in the onset of the demographic transition) in the second half of the 19th century. In particular, differences in the timing of the establishment of a national system of public education between England and Continental Europe are instrumental in isolating the role of industrial forces in human capital formation from other forces such as social control, moral conformity, enlightenment, sociopolitical stability, social and national cohesion, and military efficiency. England In the first phase of the Industrial Revolution (1760–1830), capital accumulation increased significantly without a corresponding increase in the supply of skilled labor. The investment ratio increased from 6% in 1760 to 11.7% in 1831 [Crafts (1985, p. 73)]. In contrast, literacy rates remained largely unchanged and the state devoted virtually no resources to raising the level of literacy of the masses. During the first stages of the Industrial Revolution, literacy was largely a cultural skill or a hierarchical symbol
34 This argument is supported indirectly by contemporary evidence about higher rates of return to human
capital in less developed economies than in developed economies [Psacharopoulos and Patrinos (2002)]. The greater prevalence of credit markets imperfections and other barriers for the acquisition of skills in less developed economies enabled only a partial supply response to industrial demand for human capital, contributing to this differential in the skill premium.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
207
and had limited demand in the production process.35 For instance, in 1841 only 4.9% of male workers and only 2.2% of female workers were in occupations in which literacy was strictly required [Mitch (1992, pp. 14–15)]. During this period, an illiterate labor force could operate the existing technology, and economic growth was not impeded by educational retardation.36 Workers developed skills primarily through on-the-job training, and child labor was highly valuable. The development of a national public system of education in England lagged behind other Western European countries by nearly half a century and the literacy rate hardly increased in the period 1750–1830 [Sanderson (1995, pp. 2–10)].37 As argued by Green (1990, pp. 293–294), “Britain’s early industrialization had occurred without direct state intervention and developed successfully, at least in its early stages, within a laissez-faire framework. Firstly, state intervention was thought unnecessary for developing technical skills, where the initial requirements were slight and adequately met by traditional means. Secondly, the very success of Britain’s early industrial expansion encouraged complacency about the importance of scientific skills and theoretical knowledge which became a liability in a later period when empirical knowledge, inventiveness and thumb methods were no longer adequate.” Furthermore, as argued by Landes (1969, p. 340) “although certain workers – supervisory and office personnel in particular – must be able to read and do the elementary arithmetical operations in order to perform their duties, large share of the work of industry can be performed by illiterates as indeed it was especially in the early days of the industrial revolution”. England initiated a sequence of reforms in its educational system since the 1830s and literacy rates gradually increased. The process was initially motivated by non-industrial reasons such as religion, social control, moral conformity, enlightenment, and military efficiency, as was the case in other European countries (e.g., Germany, France, Holland, Switzerland) that had supported public education much earlier. However, in light of the modest demand for skills and literacy by the capitalists, the level of governmental support was rather small.38 In the second phase of the Industrial Revolution, the demand for skilled labor in the growing industrial sector markedly increased and the proportion of children aged 5 to 14 in primary schools increased from 11% in 1855 to 25% in 1870 [Flora et al. (1983)]. Job advertisements, for instance, suggest that literacy became an increasingly desired characteristic for employment as of the 1850s [Mitch (1993, p. 292)]. In light
35 See Mokyr (1993, 2001). 36 Some have argued that the low skill requirements even declined over this period. For instance, Sanderson
(1995, p. 89) suggests that the emerging economy created a whole range of new occupations which require even less literacy and education than the old ones. 37 For instance, in his parliamentary speech in defense of his 1837 education bill, the Whig politician, Henry Brougham, reflected upon this gap: “It cannot be doubted that some legislative effort must at length be made to remove from this country the opprobrium of having done less for education of the people than any of the more civilized nations on earth” [Green (1990, pp. 10–11)]. 38 Even in 1869 the government funded only one-third of school expenditure [Green (1990, pp. 6–7)].
208
O. Galor
of the industrial competition from other countries, capitalists started to recognize the importance of technical education for the provision of skilled workers. As noted by Sanderson (1995, pp. 10–13), “reading . . . enabled the efficient functioning of an urban industrial society laced with letter writing, drawing up wills, apprenticeship indentures, passing bills of exchange, and notice and advertisement reading”. Moreover, manufacturers argued that: “universal education is required in order to select, from the mass of the workers, those who respond well to schooling and would make a good foreman on the shop floor” [Simon (1987, p. 104)]. As it became apparent that skills were necessary for the creation of an industrial society, replacing previous ideas that the acquisition of literacy would make the working classes receptive to radical and subversive ideas, capitalists lobbied for the provision of public education for the masses. The pure laissez-faire policy failed in developing a proper educational system and capitalists demanded government intervention in the provision of education. As James Kitson, a Leeds iron-master and an advocate of technical education explained to the Select Committee on Scientific Instruction (1867–1868): “. . . the question is so extensive that individual manufacturers are not able to grapple with it, and if they went to immense trouble to establish schools they would be doing it in order that others may reap the benefit” [Green (1990, p. 295)].39 An additional turning point in the attitude of capitalists towards public education was the Paris Exhibition of 1867, where the limitations of English scientific and technical education became clearly evident. Unlike the 1851 exhibition in which England won most of the prizes, the English performance in Paris was rather poor; of the 90 classes of manufacturers, Britain dominated only in 10. Lyon Playfair, who was one of the jurors, reported that: “a singular accordance of opinion prevailed that our country has shown little inventiveness and made little progress in the peaceful arts of industry since 1862”. This lack of progress “upon which there was most unanimity conviction is that France, Prussia, Austria, Belgium and Switzerland possess good systems of industrial education and that England possesses none” [Green (1990, p. 296)]. In 1868, the government established the Parliamentary Select Committee on Scientific Education. This was the origin of nearly 20 years of various parliamentary investigations into the relationship between sciences, industry, and education, that were designed to address the capitalists’ outcry about the necessity of universal public education. A sequence of reports by the committee in 1868, the Royal Commission on Scientific Instruction and the Advancement of Science during the period 1872–75, and by the Royal Commission on Technical Education in 1882, underlined the inadequate training for supervisors, managers and proprietors, as well as workers. They argued that most managers and proprietors did not understand the manufacturing process and thus, failed to promote efficiency, investigate innovative techniques, or value the skills of their workers [Green (1990, pp. 297–298)]. In particular, W.E. Forster, the Vice President of the committee of the Council of Education told The House of Commons: “Upon 39 Indeed, the Factory Act of 1802 required owners of textile mills to provide elementary instruction for their
apprentices, but the law was poorly enforced [Cameron (1989, pp. 216–217)].
Ch. 4: From Stagnation to Growth: Unified Growth Theory
209
the speedy provision of elementary education depends our industrial prosperity . . . if we leave our work-folk any longer unskilled . . . they will become overmatched in the competition of the world” [Hurt (1971, pp. 223–224)]. The reports made various recommendations which highlighted the need to redefine elementary schools, to revise the curriculum throughout the entire school system, particularly with respect to industry and manufacture, and to improve teachers’ training. In addition, in 1868, secondary schools were investigated by the Schools Inquiry Commission. It found the level of instruction in the vast majority of schools very unsatisfactory, reflecting the employment of untrained teachers and the use of antiquated teaching methods. Their main proposal was to organize a state inspection of secondary schools and to provide efficient education geared towards the specific needs of its consumers. In particular, the Royal Commission on Technical Education of 1882 confirmed that England was being overtaken by the industrial superiority of Prussia, France and the United States and recommended the introduction of technical and scientific education into secondary schools. It appears that the English government gradually yielded to the pressure by capitalists as well as labor unions, and increased its contributions to elementary as well as higher education. In the 1870 Education Act, the government assumed responsibility for ensuring universal elementary education, although it did not provide either free or compulsory education. In 1880, prior to the significant extension of the franchise of 1884 that made the working class the majority in most industrial countries, education was made compulsory throughout England. The 1889 Technical Instruction Act allowed the new local councils to set up technical instruction committees, and the 1890 Local Taxation Act provided public funds that could be spent on technical education [Green (1990, p. 299)]. School enrollment of 10-year olds increased from 40% in 1870 to 100% in 1900, the literacy rate among men, which was stable at around 65% in the first phase of the Industrial Revolution, increased significantly during the second phase, reaching nearly 100% at the end of the 19th century [Clark (2002)], and the proportion of children aged 5 to 14 in primary schools increased in the second half of the 19th century, from 11% in 1855 to 74% in 1900 [Flora et al. (1983)]. Finally, the 1902 Balfour Act marked the consolidation of a national education system. It created state secondary schools [Ringer (1979) and Green (1990, p. 6)] and science and engineering as well as their application to technology gained prominence [Mokyr (1990, 2002)]. Continental Europe The early development of public education occurred in the western countries of continental Europe (e.g., Prussia, France, Sweden, and the Netherlands) well before the Industrial Revolution. The provision of public education at this early stage was motivated by several goals such as social and national cohesion, military efficiency, enlightenment, moral conformity, sociopolitical stability as well as religious reasons. However, as was the case in England, massive educational reforms occurred in the second half of the 19th century due to the rising demand for skills in the process of industrialization. As noted by Green (1990, pp. 293–294) “In continental Europe
210
O. Galor
industrialization occurred under the tutelage of the state and began its accelerated development later when techniques were already becoming more scientific; technical and scientific education had been vigorously promoted from the center as an essential adjunct of economic growth and one that was recognized to be indispensable for countries which wished to close Britain’s industrial lead.” In France the initial development of the education system occurred well before the Industrial Revolution, but the process was intensified and transformed to satisfy industrial needs in the second phase of the Industrial Revolution. The early development of elementary and secondary education in the 17th and 18th centuries was dominated by the Church and religious orders. Some state intervention in technical and vocational training was designed to reinforce development in commerce, manufacturing and military efficiency. After the French Revolution, the state established universal primary schools. Nevertheless, enrollment rates remained rather low. The state concentrated on the development of secondary and higher education with the objective of producing an effective elite to operate the military and governmental apparatus. Secondary education remained highly selective, offering general and technical instruction largely to the middle class [Green (1990, pp. 135–137 and 141–142)]. Legislative proposals during the National Convention quoted by Cubberley (1920, pp. 514–517) are revealing about the underlying motives for education in this period: “. . . Children of all classes were to receive education, physical, moral and intellectual, best adapted to develop in them republican manners, patriotism, and the love of labor . . . They are to be taken into the fields and workshops where they may see agricultural and mechanical operations going on . . .” The process of industrialization in France, the associated increase in the demand for skilled labor, and the breakdown of the traditional apprenticeship system, significantly affected the attitude towards education. State grants for primary schools gradually increased in the 1830s and legislation made an attempt to provide primary education in all regions, extend the higher education, and provide teacher training and school inspections. The number of communities without schools fell by 50% from 1837 to 1850 and as the influence of industrialists on the structure of education intensified, education became more stratified according to occupational patterns [Anderson (1975, pp. 15, 31)]. According to Green (1990, p. 157): “[This] legislation . . . reflected the economic development of the period and thus the increasing need for skilled labor.” The eagerness of capitalists for rapid education reforms was reflected by the organization of industrial societies that financed schools specializing in chemistry, design, mechanical weaving, spinning, and commerce [Anderson (1975, pp. 86, 204)]. As was the case in England, industrial competition led industrialists to lobby for the provision of public education. The Great Exhibition of 1851 and the London Exhibition of 1862 created the impression that the technological gap between France and other European nations was narrowing and that French manufacturers ought to invest in the education of their labor force to maintain their technological superiority. Subsequently, reports on the state of the industrial education by commissions established in the years 1862 to 1865 reflected the plea of industrialists for the provision of industrial education
Ch. 4: From Stagnation to Growth: Unified Growth Theory
211
on a large scale and for the implementation of scientific knowledge in the industry. “The goal of modern education . . . can no longer be to form men of letters, idle admirers of the past, but men of science, builders of the present, initiators of the future.”40 Education reforms in France were extensive in the second phase of the Industrial Revolution, and by 1881 a universal, free, compulsory and secular primary school system had been established and technical and scientific education further emphasized. Illiteracy rates among conscripts tested at the age of 20 declined gradually from 38% in 1851–55 to 17% in 1876–80 [Anderson (1975, p. 158)], and the proportion of children aged 5 to 14 in primary schools increased from 51.5% in 1850 to 86% in 1901 [Flora et al. (1983)]. Hence, the process of industrialization, and the increase in the demand for skilled labor in the production process, led industrialists to support the provision of universal education, contributing to the extensiveness of education as well as to its focus on industrial needs. In Prussia, as well, the initial steps towards compulsory education took place at the beginning of the 18th century well before the Industrial Revolution. Education was viewed primarily as a method to unify the state. In the second part of the 18th century, education was made compulsory for all children aged 5 to 13. Nevertheless, these regulations were not strictly enforced partly due to the lack of funding (in light of the difficulty of taxing landlords for this purpose), and partly due to their adverse effect on child labor income. At the beginning of the 19th century, motivated by the need for national cohesion, military efficiency, and trained bureaucrats, the education system was further reformed. Provincial and district school boards were established, education was compulsory (and secular) for a three-year period, and the Gymnasium was reconstituting as a state institution that provided nine years of education for the elite [Cubberly (1920) and Green (1990)]. The process of industrialization in Prussia and the associated increase in the demand for skilled labor led to significant pressure for educational reforms and thereby to the implementation of universal elementary schooling. Taxes were imposed to finance the school system and teacher’s training was established. Secondary schools started to serve industrial needs as well, the Realschulen, which emphasized the teaching of mathematics and science, was gradually adopted, and vocational and trade schools were founded. Total enrollment in secondary school increased six fold from 1870 to 1911 [Flora et al. (1983)]. “School courses . . . had the function of converting the occupational requirements of public administration, commerce and industry into educational qualifications . . .” [Muller (1987, pp. 23–24)]. Furthermore, the Industrial Revolution significantly affected the nature of education in German universities. German industrialists who perceived advanced technology as the competitive edge that could boost German industry, lobbied for reforms in the operation of universities, and offered to pay to reshape their activities so as to favor their interest in technological training and industrial applications of basic research [McClelland (1980, pp. 300–301)].
40 L’enseignement professionnel (1864, p. 332), quoted in Anderson (1975, p. 194).
212
O. Galor
The structure of education in the Netherlands also reflected the interest of capitalists in the skill formation of the masses. In particular, as early as the 1830s, industrial schools were established and funded by private organizations, representing industrialists and entrepreneurs. Ultimately, in the latter part of the 19th century, the state, urged by industrialists and entrepreneurs, started to support these schools [Wolthuis (1999, pp. 92–93, 119, 139–140, 168, 171–172)]. United States The process of industrialization in the US also increased the importance of human capital in the production process. Evidence provided by Abramovitz and David (2000) and Goldin and Katz (2001) suggests that over the period 1890–1999, the contribution of human capital accumulation to the growth process of the United States nearly doubled.41 As argued by Goldin (2001), the rise of the industrial, business and commerce sectors in the late 19th and early 20th centuries increased the demand for managers, clerical workers, and educated sales personnel who were trained in accounting, typing, shorthand, algebra, and commerce. Furthermore, in the late 1910s, technologically advanced industries demanded blue-collar craft workers who were trained in geometry, algebra, chemistry, mechanical drawing, etc. The structure of education was transformed in response to industrial development and the increasing importance of human capital in the production process, and American high schools adapted to the needs of the modern workplace of the early 20th century. Total enrollment in public secondary schools increased 70-fold from 1870 to 1950 [Kurian (1994)].42 B. Human capital formation and inequality In the first phase of the Industrial Revolution, prior to the implementation of significant education reforms, physical capital accumulation was the prime engine of economic growth. In the absence of significant human capital formation, the concentration of capital among the capitalists widened wealth inequality. Once education reforms were implemented, however, the significant increase in the return to labor relative to capital, as well as the significant increase in the
41 It should be noted that literacy rates in the US were rather high prior to this increase in the demand
for skilled labor. Literacy rates among the white population were already 89% in 1870, 92% in 1890, and 95% in 1910 [Engerman and Sokoloff (2000)]. Education in earlier periods was motivated by social control, moral conformity, and social and national cohesion, as well as required skills for trade and commerce. In particular, Field (1976) and Bowles and Gintis (1975) argue that educational reforms were designed to sustain the existing social order, by displacing social problems into the school system. 42 As noted by Galor and Moav (2006), due to differences in the structure of education finance in the US in comparison to European countries, capitalists in the US had only limited incentives to lobby for the provision of education and support it financially. Unlike the central role that government funding played in the provision of public education in European countries, the evolution of the educational system in the US was based on local initiatives and funding. The local nature of the education initiatives in the US induced community members, in urban as well as rural areas, to play a significant role in advancing their schooling system. American capitalists, however, faced limited incentives to support the provision of education within a county in an environment where labor was mobile across counties and the benefits from educational expenditure in one county may be reaped by employers in other counties.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
213
Figure 29. Wealth inequality and factor prices: England 1820–1920. Sources: Williamson (1985) for inequality and Clark (2002, 2003) for factor prices.
real return to labor and the associated accumulation of assets by the workers, brought about a decline in inequality. Evidence suggests that in the first phase of the Industrial Revolution, prior to the implementation of education reforms, capital accumulation brought about a gradual increase in wages along with an increase in the wage–rental ratio. Education reforms in the second phase of the Industrial Revolution were associated with a sharp increase in real wages along with a sharp increase in the wage–rental ratio.43 Finally, wealth inequality widened in the first phase of the Industrial Revolution and reversed its course in the second phase, once significant education reforms were implemented. As documented in Figure 29, based on a controversial study, over the time period 1823–1915, wealth inequality in the UK peaked around 1870 and declined thereafter, in close association with the patterns of enrollment rates and factor prices, depicted in Figures 28 and 29.44 It appears that the decline in inequality was associated with the significant changes that occurred around 1870 in the relative returns to the main factors of production possessed by capitalists and workers. These changes in factor prices reflect the increase in enrollment rates and its delayed effect on the skill level per worker. Similar patterns of the effect of education on factor prices and, therefore, on inequality are observed in France as well. As argued by Morrisson and Snyder (2000), wealth inequality in France increased during the first half of the 19th century, and started to decline in the last decades of the 19th century in close association with the rise in education rates depicted in Figure 28, the rise in real wages depicted in Figure 10, and a 43 It should be noted that the main source of the increase in real wages was not a decline in prices. Over this
period nominal wages increased significantly as well. 44 It should be noted that the return to capital increased moderately over this period, despite the increase
in the supply of capital, reflecting technological progress, population growth, and accumulation of human capital.
214
O. Galor
declining trend of the return to capital over the 19th century. The decline in inequality in France appears to be associated with the significant changes in the relative returns to the main factors of production possessed by capitalists and workers in the second part of the 19th century. As depicted in Figure 10, based on the data presented in Levy-Leboyer and Bourguignon (1990), real wages, as well as the wage–rental ratio, increased significantly as of 1860, reflecting the rise in the demand for skilled labor and the effect of the increase in enrollment rates on the skill level per worker. The German experience is consistent with this pattern as well. Inequality in Germany peaked towards the end of the 19th century [Morrisson and Snyder (2000)] in association with a significant increase in the real wages and in the wage–rental ratio from the 1880s [Spree (1977) and Berghahn (1994)], which is in turn related to the provision of industrial education in the second half of the 19th century. The link between the expansion of education and the reduction in inequality is present in the US as well. Wealth inequality in the US, which increased gradually from colonial times until the second half of the 19th century, reversed its course at the turn of the century and maintained its declining pattern during the first half of the 20th century [Lindert and Williamson (1976)]. As argued by Goldin (2001), the emergence of the “new economy” in the early 20th century increased the demand for educated workers. The creation of publicly funded mass modern secondary schools from 1910 to 1940 provided general and practical education, contributed to workers productivity, and opened the gates for college education. This expansion facilitated social and geographic mobility and generated a large decrease in inequality in economic outcomes. C. Independence of education reforms from political reforms in the 19th century The 19th century was marked by significant political reforms along with the described education reforms. One could therefore challenge the significance of the industrial motive for education reform, suggesting that political reforms during the 19th century shifted the balance of power towards the working class and enabled workers to implement education reforms against the interest of the industrial elite. However, political reforms that took place in the 19th century had no apparent effect on education reforms over this period, strengthening the hypothesis that indeed industrial development, and the increasing demand for human capital, were the trigger for human capital formation and the demographic transition. Education reforms took place in autocratic states that did not relinquish political power throughout the 19th century, and major reforms occurred in societies in the midst of the process of democratization well before the stage at which the working class constituted the majority among the voters. In particular, the most significant education reforms in the UK were completed before the voting majority shifted to the working class. The patterns of education and political reforms in the UK during the 19th century are depicted in Figure 30. The Reform Act of 1832 nearly doubled the total electorate, but nevertheless only 13% of the votingage population were enfranchised. Artisans, the working class, and some sections of the lower middle class remained outside of the political system. The franchise was extended further in the Reform Acts of 1867 and 1884 and the total electorate nearly doubled in
Ch. 4: From Stagnation to Growth: Unified Growth Theory
215
Figure 30. The evolution of voting rights and school enrollment. Source: Flora et al. (1983).
each of these episodes. However, working-class voters did not become the majority in all urban counties until 1884 [Craig (1989)]. The onset of England’s education reforms, and in particular, the fundamental Education Act of 1870 and its major extension in 1880 occurred prior to the political reforms of 1884 that made the working class the majority in most counties. As depicted in Figure 30, a trend of significant increase in primary education was established well before the extension of the franchise in the context of the 1867 and 1884 Reform Acts. In particular, the proportion of children aged 5 to 14 in primary schools increased five-fold (and surpassed 50%) over the three decades prior to the qualitative extension of the franchise in 1884 in which the working class was granted a majority in all urban counties. Furthermore, the political reforms do not appear to have any effect on the pattern of education reform. In fact, the average growth rate of education attendance from decade to decade over the period 1855 to 1920 reaches a peak at around the Reform Act of 1884 and starts declining thereafter. It is interesting to note, however, that the abolishment of education fees in nearly all elementary schools occurs only in 1891, after the Reform Act of 1884, suggesting that the political power of the working class may have affected the distribution of education costs across the population, but the decision to educate the masses was taken independently of the political power of the working class. In France, as well, the expanding pattern of education preceded the major political reform that gave the voting majority to the working class. The patterns of education and political reforms in France during the 19th century are depicted in Figure 30. Prior to 1848, restrictions limited the electorate to less than 2.5% of the voting-age population. The 1848 revolution led to the introduction of nearly universal voting rights for males. Nevertheless, the proportion of children aged 5 to 14 in primary schools doubled (and exceeded 50%) over the two decades prior to the qualitative extension of the franchise in 1848 in which the working class was granted a majority among voters. Furthermore, the political reforms of 1848 do not appear to have any effect on the pattern of education expansion.
216
O. Galor
Figure 31. The evolution of average years of education: 1960–2000. Source: Barro and Lee (2000).
A similar pattern occurs in other European countries. Political reforms in the Netherlands did not affect the trend of education expansion and the proportion of children aged 5 to 14 in primary schools exceeded 60% well before the major political reforms of 1887 and 1897. Similarly, the trends of political and education reforms in Sweden, Italy, Norway, Prussia and Russia do not lend credence to the alternative hypothesis. Less developed economies The process of industrialization was characterized by a gradual increase in the relative importance of human capital in less developed economies as well. As depicted in Figure 31, educational attainment increased significantly across all less developed regions. Moreover, in line with the pattern that emerged among developed economies in the 19th century, the increase in educational attainment preceded or occurred simultaneously with the decline in total fertility rates. In particular, the average years of schooling in Africa increased by 44% (from 1.56 to 2.44) prior to the onset of decline in total fertility rates in 1980, as depicted in Figure 23. 2.3.4. International trade and industrialization The process of industrialization in developed economies was enhanced by the expansion of international trade. During the 19th century, North–South trade, as well as North– North trade, expanded significantly due to a rapid industrialization in Northwest Europe as well as the reduction of trade barriers and transportation costs and the benefits of the gold standard. The ratio of world trade to output was about 2% in 1800, but then it rose to 10% in 1870, to 17% in 1900 and 21% in 1913 [Estevadeordal, Frantz and Taylor (2002)]. While much of this trade occurred between industrial economies a significant proportion was between industrial and non-industrial economies. As shown in
Ch. 4: From Stagnation to Growth: Unified Growth Theory
217
Table 1 Regional shares of world trade in manufactures. Source: Yates (1959). 1876–1880
UK and Ireland Northwest Europe Other Europe U.S. and Canada Rest of the World
1896–1900
1913
Exports
Imports
Exports
Imports
Exports
Imports
37.8% 47.1% 9.2% 4.4% 1.5%
9.1% 18.1% 13.3% 7.7% 51.8%
31.5% 45.8% 10.3% 7.4% 5.0%
10.4% 20.3% 12.2% 9.6% 47.5%
25.3% 47.9% 8.3% 10.6% 7.9%
8.2% 24.4% 15.4% 12.1% 39.9%
Table 1, before 1900 nearly 50% of manufactured exports were to non-European and non-North American economies. By the end of 19th century a clear pattern of specialization emerged. The UK and Northwest Europe were net importers of primary products and net exporters of manufactured goods, whereas the exports of Asia, Oceania, Latin America and Africa were overwhelmingly composed of primary products [Findlay and O’Rourke (2003)]. Atlantic trade as well as trade with Asia, in an era of colonialism, had a major effect on European growth starting in the late 16th century [Pomeranz (2000)]. In addition, later expansion of international trade contributed further to the process of industrialization in the UK and Europe [O’Rourke and Williamson (1999)]. For the UK, the proportion of foreign trade to national income grew from about 10% in the 1780s to about 26% over the years 1837–45, and 51.5% in the time period 1909–13 [Kuznets (1967)]. Other European economies experienced a similar pattern as well. The proportion of foreign trade to national income on the eve of World War I was 53.7% in France, 38.3% in Germany, 33.8% in Italy, and 40.4% in Sweden [Kuznets (1967, Table 4)]. Furthermore, export was critical for the viability of some industries, especially the cotton industry, where 70% of the UK output was exported in the 1870s. The quantitative study of Stokey (2001) suggests that trade was instrumental for the increased share of manufacturing in total output in the UK, as well as for the significant rise in real wages, and the empirical examination of O’Rourke and Williamson (2005) demonstrates that trade was a significant force behind the rise in productivity in the UK. Thus while it appears that technological advances could have spawned the Industrial Revolution without an expansion of international trade, the growth in exports increased the pace of industrialization and the growth rate of output per capita.45
45 Pomeranz (2000), provides historical evidence for the vital role of trade in the take-off of the European
economies. He argues that technological and development differences between Europe and Asia were minor around 1750, but the discovery of the New World enabled Europe, via Atlantic trade, to overcome ‘land constraints’ and to take off technologically.
218
O. Galor
Figure 32. The great divergence. Source: Maddison (2001).
2.4. The great divergence The differential timing of the take-off from stagnation to growth across countries, and the corresponding variations in the timing of the demographic transition, led to a great divergence in income per capita as well as population growth. The last two centuries have witnessed dramatic changes in the distribution of income and population across the globe. Some regions have excelled in the growth of income per capita, while other regions have been dominant in population growth. Inequality in the world economy was negligible till the 19th century. The ratio of GDP per capita between the richest region and the poorest region in the world was only 1.1 : 1 in 1000, 2 : 1 in 1500 and 3 : 1 in 1820. As depicted in Figure 32, there has been a ‘Great Divergence’ in income per capita among countries and regions in the past two centuries. In particular, the ratio of GDP per capita between the richest group (Western Offshoots) and the poorest region (Africa) has widened considerably from a modest 3 : 1 ratio in 1820, to 5 : 1 ratio in 1870, 9 : 1 ratio in 1913, 15 : 1 in 1950, and 18 : 1 ratio in 2001. An equally momentous transformation occurred in the distribution of world population across regions, as depicted in Figure 33. The earlier take-off of Western European countries increased the amount of resources that could be devoted for the increase in family size, permitting a 16% increase in the share of their population in the world economy within a 50 year period (from 12.8% in 1820 to 14.8% in 1870). However, the early onset in the Western European demographic transition and the long delay in the demographic transition of less developed regions, well into the 2nd half of the 20th century, led to a 55% decline in the share of Western European population in the world, from 14.8% in 1870 to 6.6% in 1998. In contrast, the prolongation of the Post-Malthusian period among less developed regions, in association with the delay in their demographic transition well into the second half of 20th century, channeled their increased resources towards a significant increase in their population. Africa’s share
Ch. 4: From Stagnation to Growth: Unified Growth Theory
219
Figure 33. Divergence in regional populations. Source: Maddison (2001).
of world population increased 84%, from 7% in 1913 to 12.9% in 1998, Asia’s share of world population increased 11% from 51.7% in 1913 to 57.4% in 1998, and Latin American countries increased their share in world population from 2% in 1820 to 8.6% in 1998. Thus, while the ratio of income per capita in Western Europe to that in Asia has tripled in the last two centuries, the ratio of Asian to European population has doubled.46 The divergence that has been witnessed in the last two centuries has been maintained across countries in the last decades as well [e.g., Jones (1997) and Pritchett (1997)]. Interestingly, however, Sala-i-Martin (2002) shows that divergence has not been observed in recent decades across people in the world (i.e., when national boundaries are removed).
3. The fundamental challenges The establishment of a unified theory of economic growth that can account for the intricate process of development over the course of the last thousands of years has been one of the most significant research challenges faced by researchers in the field of growth and development. A unified theory unveils the underlying micro-foundations that are consistent with the entire process of economic development, enhancing the confidence in the viability of growth theory, its predictions and policy implications, while improving the understanding of the driving forces that led to the recent transition from stagnation to growth and the Great Divergence. Moreover, a comprehensive 46 Over the period 1820–1998, the ratio between income per capita in Western Europe and Asia (excluding
Japan) grew 2.9 times, whereas the ratio between the Asian population (excluding Japan) and the Western European population grew 1.7 times [Maddison (2001)].
220
O. Galor
understanding of the hurdles faced by less developed economies in reaching a state of sustained economic growth would be futile, unless the forces that initiated the transition of the currently developed economies into a state of sustained economic growth would be identified, and modified, to account for the differences in the evolutionary structure of less developed economies in an interdependent world. The evidence presented in Section 2 suggests that the preoccupation of growth theory with the empirical regularities that have characterized the growth process of developed economies in the past century and of less developed economies in the last few decades, has become harder to justify from a scientific viewpoint. Could we justify the use of selective observations about the recent course of the growth process and its principal causes in the formulation of exogenous and endogenous growth models? Could we be confident about the predictions of a theory that is not based on micro-foundations that match the major characteristics of the entire growth process? The evolution of theories in older scientific disciplines suggests that theories that are founded on the basis of a subset of the existing observations are fragile and non-durable, and are often generating increasingly distorted predictions. 3.1. Mysteries of the growth process The underlying determinants of the stunning recent escape from the Malthusian trap have been shrouded in mystery and their significance for the understanding of the contemporary growth process has been explored only very recently. What are the major economic forces that led to the epoch of Malthusian stagnation that characterized most of human history? What is the origin of the sudden spurt in growth rates of output per capita and population that occurred in the course of the take-off from stagnation to growth? Why had episodes of technological progress in the pre-industrialization era failed to generate sustained economic growth? What was the source of the dramatic reversal in the positive relationship between income per capita and population that existed throughout most human history? What are the main forces that prompted the demographic transition? Would the transition to a state of sustained economic growth be feasible without the demographic transition? Are there underlying unified behavioral and technological structures that can account for these distinct phases of development simultaneously and what are their implications for the contemporary growth process? The mind-boggling phenomenon of the Great Divergence in income per capita across regions of the world in the past two centuries, that accompanied the take-off from an epoch of stagnation to a state of sustained economic growth, presents additional unresolved mysteries about the growth process. What accounts for the sudden take-off from stagnation to growth in some countries in the world and the persistent stagnation in others? Why has the positive link between income per capita and population growth reversed its course in some economies but not in others? Why have the differences in income per capita across countries increased so markedly in the last two centuries? Did the pace of transition to sustained economic growth in advanced economies adversely affect the process of development in less-developed economies?
Ch. 4: From Stagnation to Growth: Unified Growth Theory
221
The transitions from a Malthusian epoch to a state of sustained economic growth and the emergence of the Great Divergence have shaped the current growth process in the world economy. Nevertheless, non-unified growth models overlooked these significant underlying forces of the process of development. 3.2. The incompatibility of non-unified growth theories Existing (non-unified) growth models are unable to capture the growth process over the entire course of human history. Malthusian models capture the growth process during the Malthusian epoch, but are incompatible with the transition to the Modern Growth Regime. Neoclassical growth models (with endogenous or exogenous technological change), in contrast, are compatible with the growth process of the developed economies during the Modern Growth Regime, but fail to capture the evolution of economies during the Malthusian epoch, the origin of the take-off from the Malthusian epoch into the Post-Malthusian Regime, and the sources of the demographic transition and the emergence of the Modern Growth Regime. Moreover, the failure of non-unified growth models in identifying the underlying factors that led to the transition from stagnation to growth, limits their applicability for the contemporary growth process of the less developed economies, and thereby for the current evolution of the world income distribution. 3.2.1. Malthusian and Post-Malthusian theories The Malthusian theories The Malthusian theory, as was outlined initially by Malthus (1798), captures the main attributes of the epoch of Malthusian stagnation that had characterized most of human existence, but is utterly inconsistent with the prime characteristics of the Modern Growth Regime.47 The theory suggests that the stagnation in the evolution of income per capita over this epoch reflected the counterbalancing effect of population growth on the expansion of resources, in an environment characterized by diminishing returns to labor. The expansion of resources, according to Malthus, led to an increase in population growth, reflecting the natural result of the “passion between the sexes”.48 In contrast, when population size grew beyond the capacity of the available resources, it was reduced by the “preventive check” (i.e., intentional reduction of fertility) as well as by the “positive check” (i.e., the tool of nature due to malnutrition, disease, and famine). According to the theory, periods marked by the absence of changes in the level of technology or in the availability of land, were characterized by a stable population size 47 The Malthusian theory was formalized recently. Kremer (1993) models a reduced-form interaction be-
tween population and technology along a Malthusian equilibrium, and Lucas (2002) presents a Malthusian model in which households optimize over fertility and consumption, labor is subjected to diminishing returns due to the presence of a fixed quantity of land, and the Malthusian level of income per capita is determined endogenously. 48 As argued by Malthus (1798), “The passion between the sexes has appeared in every age to be so nearly the same, that it may always be considered, in algebraic language as a given quantity.”
222
O. Galor
as well as a constant income per capita. In contrast, episodes of technological progress, land expansion, and favorable climatic conditions, brought about temporary gains in income per capita, triggering an increase in the size of the population which led eventually to a decline in income per capita to its long-run level. The theory proposes, therefore, that variation in population density across countries during the Malthusian epoch reflected primarily cross-country differences in technologies and land productivity. Due to the positive adjustment of population to an increase in income per capita, differences in technologies or in land productivity across countries resulted in variations in population density rather than in the standard of living. The Malthusian theory generates predictions that are largely consistent with the characteristics of economies during the Malthusian epoch, as described in Section 2.1. It suggests that: (a) technological progress or resource expansion would lead to a larger population, without altering the level of income in the long run, (b) income per capita would fluctuate during the Malthusian epoch around a constant level, and (c) technologically superior countries would have eventually denser populations but their standard of living in the long run would not reflect the degree of their technological advancement. These predictions, however, are irremediably inconsistent with the relationship between income per capita and population that has existed in the post-demographic transition era as well as with the state of sustained economic growth that had characterized the Modern Growth Regime. Unified theories of economic growth, in contrast, incorporate the main ingredients of the Malthusian economy into a broader context, focusing on the interaction between technology, the size of the population, and the distribution of its characteristics, generating the main ingredients of the Malthusian epoch as well as an inevitable take-off to the Post Malthusian Regime and the Modern Growth Regime. The Post-Malthusian theories The Post-Malthusian theories capture the acceleration of the growth rate of income per capita and population growth that occurred during the Post-Malthusian Regime in association with the process of industrialization. They do not capture, however, the stagnation during the Malthusian epoch and the economic forces that gradually emerged in this era and brought about the take-off from the Malthusian trap. Moreover, these theories do not account for the factors that ultimately triggered the demographic transition and the shift to a state of sustained economic growth.49
49 Models that are not based on Malthusian elements are unable to capture the long epoch of Malthusian stag-
nation in which output per capita fluctuates around a subsistence level. For instance, an interesting research by Goodfriend and McDermott (1995) demonstrates that exogenous population growth increases population density and hence generates a greater scope for the division of labor inducing the development of markets and economic growth. Their model, therefore, generates a take-off from non-Malthusian stagnation to Post-Malthusian Regime in which population and output are positively related. The model lacks Malthusian elements and counter-factually it implies, therefore, that since the emergence of a market economy over 5000 years ago growth has been strictly positive. Moreover, it does not generate the forces that would bring about
Ch. 4: From Stagnation to Growth: Unified Growth Theory
223
These theories suggest that during the Post-Malthusian Regime the acceleration in technological progress and the associated rise in income per capita was only channeled partly towards an increase in the size of the population. Although, the Malthusian mechanism, linking higher income to higher population growth, continued to function, the effect of higher population on the dilution of resources per capita, and thus on the reduction of income per capita, was counteracted by the acceleration in technological progress and capital accumulation, allowing income per capita to rise despite the offsetting effects of population growth. Kremer (1993), in an attempt to defend the role of the scale effect in endogenous growth models, examines a reduced-form of the co-evolution of population and technology in a Malthusian and Post Malthusian environment, providing evidence for the presence of a scale effect in the pre-demographic transition era.50 Kremer’s Post-Malthusian theory, however, does not identify the factors that brought about the take-off from the Malthusian trap, as well as the driving forces behind the demographic transition and the transition to a state of sustained economic growth. Unified theories capture the main characteristics of the Post-Malthusian Regime, and generate in contrast, the endogenous driving forces that brought about the take-off from the Malthusian epoch into this regime and ultimately enabled the economy to experience a demographic transition and to reside in a state of sustained economic growth. 3.2.2. Theories of modern economic growth Exogenous growth models [e.g., Solow (1956)] that have focused primarily on the role of factor accumulation in the growth process, as well as endogenous growth models [e.g., Romer (1990), Grossman and Helpman (1991) and Aghion and Howitt (1992)] that have devoted their attention to the role of endogenous technological progress in the process of development, were designed to capture the main characteristics of the Modern Growth Regime. These models, however, are inconsistent with the pattern of development that had characterized economies over most of human history, and they do not posses the research methodology that could shed light on the process of development in its entirety. Non-unified growth models do not unveil the underlying micro-foundations of the intricate patterns of the growth process over human history, and thus they could not capture the epoch of Malthusian stagnation that characterized most of human history, the underlying driving forces that triggered the transition from stagnation to growth, the hurdles faced by less developed economies in reaching a state of sustained economic
the demographic transition and ultimately sustained economic growth. In the long-run the economy remains in the Post-Malthuisan Regime in which the growth of population and output are positively related. Other non-Malthusian models that abstract from population growth and generate an acceleration of output growth along the process of industrialization include Acemoglu and Zilibotti (1997). 50 Komlos and Artzrouni (1990) simulate an escape from a Malthusian trap based on the Malthusian and Boserupian interaction between population and technology.
224
O. Galor
growth, and the associated phenomenon of the Great Divergence in income per capita across countries.51 Moreover, although the evolution of the demographic regime in the course of human history appears essential for the understanding of the evolution of income per capita over the process of development, most endogenous and exogenous growth models abstract from the determination of population growth over the growth process, and their predictions are inconsistent with the demographic structure over the course of human history.52 Non-unified growth models with endogenous population have been largely oriented toward the modern regime, capturing some aspects of the recent negative relationship between population growth and income per capita, but failing to capture the significance of the positive effect of income per capita on population growth that had characterized most of human existence, as well as the economic factors that triggered the demographic transition and the take-off to a state of sustained economic growth.53 3.3. Theories of the demographic transition and their empirical assessment This section examines various mechanisms that have been proposed as possible triggers for the demographic transition, assessing their empirical validity, and their potential role in the transition from stagnation to growth.54 The demographic transition that swept the world in the course of the last century has been identified as one of the prime forces in the movement from an epoch of stagnation to a state of sustained economic growth, enabling economies to convert a larger share of the fruits of factor accumulation and technological progress into growth of income per capita. Theories of the demographic transition attempt to capture the determinants of this significant reduction in fertility rates and population growth in various regions of
51 As long as the neoclassical production structure of non-decreasing returns to scale is maintained, non-
unified growth models could not be modified to account for the Malthusian epoch by the incorporation of endogenous population growth. For instance, suppose that the optimal growth model would be augmented to account for endogenous population. Suppose further, that the parameters of the model would be chosen so as to assure that the level of income per capita would reflect the level that existed during the Malthusian epoch and population growth will be near replacement level as was the case during this era. This equilibrium would not possess the prime characteristic of a Malthusian equilibrium. Namely, technological progress would raise income per capita permanently due to the fact that adjustments in population growth would not offset this rise of income (as long as the return to labor is characterized by non-diminishing returns to scale). 52 In fact, most endogenous growth models that focus exclusively on the modern growth regime are inconsistent with the demographic structure within this regime, predicting a positive effect of population growth on (the growth rate of) income per capita. A notable exception is Dalgaard and Kreiner (2001). 53 Research that capture aspects of the cross-section relationship between income per capita and fertility include Razin and Ben-Zion (1975), Barro and Becker (1989) and Becker, Murphy and Tamura (1990), and more recently Dahan and Tsiddon (1998), Kremer and Chen (2002), McDermott (2002), De la Croix and Doepke (2003), and Moav (2005). 54 See Galor (2005) as well.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
225
the world in the past century, in the aftermath of an unprecedented increase in population growth during the Post-Malthusian Regime. The simultaneity of the demographic transition across Western European countries provides a fertile ground for the examination of the validity of the various theories in the context of countries that appear in similar stages of development, and are not overly diverse in their sociocultural heritage. The simultaneous reversal in the significant upward trend in fertility rates among Western European countries suggests that a common economic force may have triggered the demographic transition in this region and is likely to be the driving force behind the onset of the fertility decline in other regions of the world as well. Was the onset of the demographic transition across Western European countries an outcome of a simultaneous decline in mortality rates? Was it associated with the higher levels of income enjoyed by Western European countries in the process of industrialization? Was it an outcome of the rise in the relative wages of women in the second phase of the Industrial Revolution? Or, an outcome of the regional acceleration in technological progress and its impact on the universal rise in the industrial demand for human capital in the second phase of the Industrial Revolution? Historical evidence suggests that demographers’ preferred explanation for the demographic transition – the decline in mortality rates – does not account for the reversal of the positive historical trend between income and fertility. Moreover, the role attributed to higher income levels in the demographic transition appears implausible. The evidence suggests that the rise in the demand for human capital is the most significant force behind the demographic transition, and it is therefore a central building block in unified growth theory.
3.3.1. The decline in infant and child mortality The decline in infant and child mortality rates that preceded the decline in fertility rates in many countries in the world, with the notable exceptions of France and the US, has been demographers’ favorite explanation for the onset of the decline in fertility in the course of the demographic transition.55 Nevertheless, it appears that this viewpoint is based on weak theoretical reasoning and is inconsistent with historical evidence. While it is highly plausible that mortality rates were among the factors that affected the level of total fertility rates along human history, historical evidence does not lend credence to the argument that the decline in mortality rates accounts for the reversal of the positive historical trend between income and fertility. The decline in mortality rates does not appear to be the trigger for the decline in fertility in Western Europe. As demonstrated in Figures 22 and 24, the mortality decline in Western Europe started nearly a century prior to the decline in fertility and it was 55 The effect of the decline in mortality rates on the prolongation of productive life and thus, on the return to
human capital is discussed in Section 3.3.3.
226
O. Galor
Figure 34. Investment in human capital and the demographic transition, England, 1730–1935. Sources: Flora et al. (1983) and Wrigley and Schofield (1981).
associated initially with increasing fertility rates in some countries and non-decreasing fertility rates in other countries. In particular, as demonstrated in Figure 34, the decline in mortality started in England in the 1730s and was accompanied by a steady increase in fertility rates until 1820.56 The rise in income per capita in the Post-Malthusian Regime increased the desirable number of surviving offspring and thus, despite the decline in mortality rates, fertility did not fall so as to reach this higher desirable level of surviving offspring.57 As depicted in Figure 34, the decline in fertility during the demographic transition occurred in a period in which the pattern of declining mortality (and its adverse effect on fertility) maintained the trend that existed in the 140 years that preceded the demographic transition.58 The reversal in the fertility patterns in England as well as other Western European countries in the 1870s suggests, therefore, that the demographic transition was prompted by a different universal force than the decline in infant and child mortality – a force that reflects a significant change in course prior to the demographic transition. Furthermore, most relevant from an economic point of view is the cause of the reduction in net fertility (i.e. the number of children reaching adulthood). The decline in 56 As documented by Chesnais (1992), the evolutionary patterns of infant mortality rates and crude birth rates
were rather similar. 57 The same theoretical reasoning is applicable for countries in which fertility rates remained stable over this
period. 58 One could argue that the decline in mortality was not internalized into the decision of households who had
difficulties separating temporary decline from a permanent one. This argument is highly implausible given the fact that mortality declined monotonically for nearly 140 years prior to the demographic transition. It is inconceivable that six generations of households did not update information about mortality rates in their immediate surrounding, while keeping the collective memories about mortality rates two centuries earlier.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
227
the number of surviving offspring that was observed during the demographic transition (e.g., Figure 22) is unlikely to follow from mortality decline. Mortality decline would lead to a reduction in the number of surviving offspring if the following conditions would be satisfied:59 (i) there exists a precautionary demand for children, i.e., individuals are significantly risk averse with respect to the number of their surviving offspring and thus they hold a buffer stock of children in a high mortality environment – highly improbable from an evolutionary perspective, (ii) risk aversion with respect to consumption is not larger than risk aversion with respect to fertility – evolutionary theory would suggest the opposite), (iii) sequential fertility (i.e., replacement of non-surviving children) is modest,60 and (iv) parental resources saved from the reduction in the number of children that do not survive to adulthood are not channeled towards childbearing.61 The quantitative analysis of Doepke (2005) supports the viewpoint that a decline in infant mortality rates was not the trigger for the decline in net fertility during the demographic transition. Utilizing the mortality and fertility data from England in the time period 1861–1951, he shows that in the absence of changes in other factors, the decline in child mortality in this time period should have resulted in a rise in net fertility rates, in contrast to the evidence. Similar conclusions about the insignificance of the mortality decline for the decline in fertility during the demographic transition is reached in the quantitative analysis of Fernandez-Villaverde (2005). 3.3.2. The rise in the level of income per capita The rise in income per capita prior to the demographic transition has led some researchers to argue that the demographic transition was triggered by the rise in income per capita and its asymmetric effects on the income of households on the one hand and the opportunity cost of raising children on the other hand. Becker (1981) advanced the argument that the decline in fertility in the course of the demographic transition is a by-product of the rise in income per capita that preceded the demographic transition. He argues that the rise in income induced a fertility decline because the positive income effect on fertility that was generated by the rise in wages was dominated by the negative substitution effect that was brought about by the rising 59 In particular, the theoretical analysis of Kalemli-Ozcan (2002) generates a reduction in net fertility in
reaction to a decline in mortality assuming (implicitly) that all these conditions are satisfied. Eckstein et al. (1999) argue in their structural quantitative analysis of the demographic transition in Sweden, that mortality decline played a role in the demographic transition. Their underlying theoretical structure, however, requires conditions (iii) and (iv) as well as specific interactions between mortality, wages, and the return to human capital. 60 Doepke (2005) shows that regardless of the degree of risk aversion, the feasibility of sequential fertility is sufficient to preclude the decline in net fertility in reaction to a decline in mortality. 61 An additional force that operates against the decline in the number of surviving offspring, as a result of mortality decline, is the physiological constraint on the feasible number of birth per woman. If this constraint is binding for some households in a high mortality regime, a reduction in mortality would operate towards an increase in the number of surviving offspring.
228
O. Galor
opportunity cost of children. Similarly, Becker and Lewis (1973) argue that the income elasticity with respect to child quality is greater than that with respect to child quantity, and hence a rise in income led to a decline in fertility along with a rise in the investment in each child. This theory appears counter-factual. It suggests that the timing of the demographic transition across countries would reflect differences in income per capita. However, remarkably, as depicted in Figure 22, the decline in fertility occurred in the same decade across Western European countries that differed significantly in their income per capita. In 1870, on the eve of the demographic transition, England was the richest country in the world, with a GDP per capita of $3191.62 In contrast, Germany that experienced the decline in fertility in the same years as England, had in 1870 a GDP per capita of only $1821 (i.e., 57% of that of England). Sweden’s GDP per capita of $1664 in 1870 was 48% of that of England, and Finland’s GDP per capita of $1140 in 1870 was only 36% of that of England, and nevertheless, their demographic transitions occurred in the same decade as well.63 The simultaneity of the demographic transition across Western European countries that differed significantly in their income per capita suggests that the high level of income that was reached by these countries in the Post-Malthusian Regime had a limited role in the demographic transition. Furthermore, cross-section evidence within countries suggests that the elasticity of the number of surviving offspring with respect to income was positive prior to the demographic transition [e.g., Clark and Hamilton (2003)], in contrast to Becker’s argument that would require, at least at high income levels, a negative relationship. Moreover, a quantitative analysis of the demographic transition in England, conducted by Fernandez-Villaverde (2005), demonstrates that Becker’s theory is counter-factual. In contrast to Becker’s theory, the calibration suggests that a rise in income would have resulted in an increase in fertility rates, rather than in the observed decline in fertility. Interestingly, despite the large differences in the levels of income per capita across European countries that experienced the demographic transition in the same time period, the growth rates of income per capita of these countries were rather similar, ranging from 1.9% per year in the UK, 2.12% in Norway, 2.17% in Sweden, and 2.87% in Germany, over the period 1870–1913. This observation is consistent with theories that underlined the critical role of the acceleration in technological progress, via its effect on the industrial demand for human capital, on the onset of the demographic transition [e.g., Galor and Weil (2000) and Galor and Moav (2002)].
62 Source: Maddison (2001). GDP per capita is measured in 1990 international dollars. 63 One can argue that the income threshold for the domination of the substitution effect differ across these
set of countries due to sociocultural factors. However, the likelihood that these differential thresholds were reached within the same decade appears remote.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
229
3.3.3. The rise in the demand for human capital The rise in the demand for human capital in the second phase of industrialization of less developed economies, as documented in Section 2.3.3, and its close association with the timing of the demographic transitions, has led researchers to argue that the increasing role of human capital in the production process induced households to increase their investment in the human capital of their offspring, leading to the onset of the demographic transition. Galor and Weil (1999, 2000) argue that the acceleration in the rate of technological progress gradually increased the demand for human capital in the second phase of the Industrial Revolution, inducing parents to invest in the human capital of their offspring.64 The increase in the rate of technological progress and the associated increase in the demand for human capital brought about two effects on population growth. On the one hand, improved technology eased households’ budget constraints and provided more resources for quality as well as quantity of children. On the other hand, it induced a reallocation of these increased resources toward child quality. In the early stages of the transition from the Malthusian Regime, the effect of technological progress on parental income dominated, and population growth as well as the average population quality increased. Ultimately, further increases in the rate of technological progress, that were stimulated by human capital accumulation, induced a reduction in fertility rates, generating a demographic transition in which the rate of population growth declined along with an increase in the average level of education. Thus, consistent with historical evidence, the theory suggests that prior to the demographic transition, population growth increased along with investment in human capital, whereas the demographic transition brought about a decline in population growth along with a further increase in human capital formation.65 Galor and Weil’s theory suggests that a universal rise in the demand for human capital in the second phase of the Industrial Revolution and the simultaneous increase in educational attainment across Western European countries generated the observed simultaneous onset of the demographic transition across Western European countries that differed significantly in their levels of income per capita. The rise in the demand for human capital in the second phase of the Industrial Revolution (as documented in Section 2.3.3) led to a significant increase in the investment in children’s education and therefore to a decline in fertility. In particular, as depicted in Figure 34, the demographic transition in England was associated with a significant increase in the investment in child quality as reflected by 64 The effect of an increase in the return to human capital on parental choice of quantity and quality of
offspring is discussed in Becker (1981). 65 Quantitative evidence provided by Greenwood and Seshadri (2002) is supportive of the role of the rise in
the demand for skilled labor in the demographic transition in the US. They demonstrate that faster technological progress in an industrial skilled-intensive sector, than that in an unskilled-intensive agricultural sector, generates a demographic pattern that matches the data on the US demographic transition.
230
O. Galor
years of schooling. Quantitative evidence provided by Doepke (2004) suggests that indeed educational policy that promoted human capital formation played an important role in the demographic transition in England. Reinforcing mechanisms The decline in child labor The effect of the rise in the industrial demand for human capital on the reduction in the desirable number of surviving offspring was magnified via its adverse effect on child labor. It gradually increased the wage differential between parental labor and child labor, inducing parents to reduce the number of their children and to enhance the investment in their quality [Hazan and Berdugo (2002)].66 Moreover, the rise in the importance of human capital in the production process induced industrialists to support education reforms [Galor and Moav (2006)] and thus laws that abolish child labor [Doepke and Zilibotti (2005)], inducing a reduction in the prevalence of child labor and thus in fertility. Doepke (2004) provides quantitative evidence that suggests that indeed child labor law, and to a lesser extent educational policy, played an important role in the demographic transition in England. The rise in life expectancy The impact of the increase in the demand for human capital on the decline in the desirable number of surviving offspring may have been reinforced by the rise in life expectancy. The decline in mortality rates in developed countries since the 18th century, as depicted in Figure 24, and the recent decline in mortality rates in less developed countries, as depicted in Figure 25, corresponded to a gradual increase in life expectancy. As depicted in Figure 27, life expectancy increased in Western Europe during the 19th century from 36 in 1820 to 46 in 1900, 67 in 1950, and 78 in 1999. In particular, as depicted in Figure 26, life expectancy in England increased at a stable pace from 32 years in the 1720s to about 41 years in the 1870. This pace subsequently increased and life expectancy reached 50 years in 1900, 69 years in 1950, and 77 years in 1999. In less developed economies, in contrast, life expectancy increased markedly in the 20th century. Despite the gradual rise in life expectancy prior to the demographic transition, investment in human capital was rather insignificant as long as a technological demand for human capital did not emerge. In particular, the increase in life expectancy in England occurred 150 years prior to the demographic transition and may have resulted in a gradual increase in literacy rates, but not at a sufficient level to induce a reduction in fertility. Similarly, the rise in life expectancy in less developed regions in the first half of
66 The hypothesis of Hazan and Berdugo (2002) is consistent with existing historical evidence. For instance,
Horrell and Humphries (1995) suggest, based on data from the United Kingdom, that earnings of children age 10–14 as a percentage of father’s earning, declined from the period 1817–1839 to the period 1840–1872 by nearly 50% if the father was employed in a factory. Interestingly, the effect is significantly more pronounced if the father was employed in skilled occupations, reflecting the rise in the relative demand for skilled workers and its effect on the decline in the relative wages of children.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
231
the 20th century has not generated a significant increase in education and a demographic transition. In light of the technologically-based rise in the demand for human capital in the second phase of the Industrial Revolution, however, the rise in the expected length of productive life has increased the potential rate of return to investments in children’s human capital, and thus re-enforced and complemented the inducement for investment in education and the associated reduction in fertility rates.67 Changes in marriage institutions The effect of the rise in the demand for human capital on the desirable quality of children, and thus on the decline in fertility was reinforced by changes in marriage institutions. Gould, Moav and Simhon (2003) suggest that the rise in the demand for human capital increased the demand for educated women who have a comparative advantage in raising quality children, increasing the cost of marriage. Polygamy therefore became less affordable, inducing the transition from polygamy to monogamy, and reinforcing the decline in fertility. Edlund and Lagerlof (2002) suggest that love marriage, as opposed to arranged marriage, redirected the payment for the bride from the parent to the couple, inducing investment and human capital accumulation and thus reinforcing the decline in fertility. Natural selection and the evolution of preferences for offspring’s quality The impact of the increase in the demand for human capital on the decline in the desirable number of surviving offspring may have been magnified by cultural or genetic evolution in the attitude of individuals toward child quality. An evolutionary change in the attitude of individuals towards human capital could have generated a swift response to the increase in demand for human capital, generating a rapid decline in fertility along with an increase in human capital formation. Human beings, like other species, confront the basic trade-off between offspring’s quality and quantity in their implicit Darwinian survival strategies. Preferences for child quantity as well as for child quality reflect the well-known variety in the quantity-quality survival strategies (the K and r strategies) that exists in nature [e.g., MacArthur and Wilson (1967)]. Moreover, the allocation of resources between offspring quantity and quality is subjected to evolutionary changes [Lack (1954)]. 67 This mechanism was outlined by Galor and Weil (1999) and was examined in different settings by Erlich
and Lui (1991), Soares (2005), and Hazan and Zoabi (2004). It should be noted, however, that as argued by Moav (2005), the rise in the potential return to investment in child quality due to prolongation of the productive life is not as straightforward as it may appear. It requires that the prolongation of life would affect the return to quality more than the return to quantity. For example, if parents derive utility from the aggregate wage income of their children, prolongation of life would increase the return to quantity and quality symmetrically. Hence, an additional mechanism that would increase the relative complementarity between life expectancy and human capital would be needed to generate the rise in the return to child quality. For instance, Hazan and Zoabi (2004) suggest that an increase in life expectancy, and thus the health of students, enhances the production process of human capital and thus increases the relative return to child quality. Alternatively, Moav (2005) argues that an increase in life expectancy, while having no effect on parental choice between quality and quantity, induces the offspring to increase their own human capital, lowering fertility rates in the next generation due to the comparative advantage of educated parents in educating their children.
232
O. Galor
Galor and Moav (2002) propose that during the epoch of Malthusian stagnation that characterized most of human existence, individuals with a higher valuation for offspring quality gained an evolutionary advantage and their representation in the population gradually increased. The agricultural revolution facilitated the division of labor and fostered trade relationships across individuals and communities, enhancing the complexity of human interaction and raising the return to human capital. Moreover, the evolution of the human brain in the transition to Homo sapiens and the complementarity between brain capacity and the reward for human capital increased the evolutionary optimal investment in the quality of offspring. The distribution of valuation for quality lagged behind the evolutionary optimal level and offspring of individuals with traits of higher valuation for their offspring’s quality generated higher income and, in the Malthusian epoch when income was positively associated with aggregate resources allocated to child rearing, a larger number of offspring. Thus, the trait of higher valuation for quality gained the evolutionary advantage, and the Malthusian pressure gradually increased the representation of individuals whose preferences were biased towards child quality.68 This evolutionary process was reinforced by its interaction with economic forces. As the fraction of individuals with high valuation for quality increased, technological progress intensified, raising the rate of return to human capital. The increase in the rate of return to human capital along with the increase in the bias towards quality in the population reinforced the substitution towards child quality, setting the stage for a more rapid decline in fertility along with a significant increase in investment in human capital. 3.3.4. The decline in the gender gap The rise in women’s relative wages in the process of development, and its potential impact on the rise in female labor force participation and the associated decline in fertility rates, have been the center of another theory of the demographic transition that generates the observed hump-shaped relationship between income per capita and population growth, as depicted in Figure 11. The rise in women’s relative wages along with declining fertility rates have been observed in a large number of developed and less developed economies. In particular, as depicted in Figure 35, this pattern is observed in the US during the period 1800–1940. Galor and Weil (1996) argue that technological progress and capital accumulation increased the relative wages of women in the process of industrialization, triggering the onset of the demographic transition. They maintain that technological progress along with physical capital accumulation complemented mental-intensive tasks rather than
68 As discussed in Section 5, this mechanism is consistent with the gradual rise in literacy rates prior to the
Industrial Revolution, as depicted in Figure 42. It suggests that the increase in the investment in human capital prior to the Industrial Revolution was a reflection of changes in the composition of preference for quality in the population that stimulated investment in human capital, prior to the increase in the demand for human capital in the second phase of the Industrial Revolution.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
233
Figure 35. Female relative wages and fertility rates, United States 1800–1990. Sources: U.S. Bureau of the Census (1975) and Hernandez (2000).
physical-intensive tasks (i.e., brain rather than brawn) and thus, in light of the comparative advantage of women in mental-intensive tasks, the demand for women’s labor input gradually increased in the industrial sector, increasing the absolute wages of men and women but decreasing the gender wage gap.69 As long as the rise in women’s wages was insufficient to induce a significant increase in women’s labor force participation, fertility increased due to the rise in men’s wages.70 Ultimately, however, the rise in women’s relative wages was sufficient to induce a significant increase in labor force participation, generating a demographic transition. Unlike the single-parent model in which an increase in income generates conflicting income and substitution effects that cancel one another, if preferences are homothetic, in the two-parent household model, if most of the burden of child rearing is placed on women, a rise in women’s relative wages increases the opportunity cost of raising children more than household’s income, generating a pressure towards a reduction in fertility. 69 See Goldin (1990) as well. 70 The U-shaped pattern of female labor force participation in the process of industrialization follows from
the coexistence of an industrial sector and a non-modern production sector that is not fully rival with child rearing. Women’s marginal product in the non-modern sector was not affected by capital accumulation in the industrial sector, while women’s potential wages in the modern sector increased. In the early process of industrialization, therefore, capital accumulation increased labor productivity in the industrial sector, family income increased via men’s wages, while female wages, based on the production of market goods in the home sector did not change. Fertility increased due to the income effect generated by the rise in men’s wages, and female labor force participation fell. Eventually, capital accumulation and technological progress in the industrial sector increased female wages sufficiently, inducing a rise in female labor force participation in the industrial sector and reducing fertility.
234
O. Galor
Figure 36. The decline in the human capital gap between male and female: England 1840–1900. Source: Cipolla (1969).
The process of development was associated, in addition, with a gradual decline in the human capital gap between male and female. As depicted in Figure 36, literacy rates among women which were only 76% of those among men in 1840, grew faster in the 19th century reaching men’s level in 1900. The rise in the demand for human capital in the process of development induced a gradual improvement in the level of female education, raising the opportunity cost of children more than household’s income, and triggering a fertility decline [Lagerlof (2003b)]. 3.3.5. Other theories The old-age security hypothesis The old-age security hypothesis has been proposed as an additional mechanism for the onset of the demographic transition. It suggests that, in the absence of capital markets that permit intertemporal lending and borrowing, children are assets that permit parents to smooth consumption over their lifetime.71 Hence, the process of development and the establishment of capital markets reduced this motivation for rearing children, contributing to the demographic transition. Although old-age support is a plausible element that may affect the level of fertility, it appears as a minor force in the context of the demographic transition. First, since there are rare examples in nature of offspring that support their parents in old age, it appears that old-age support could not be the prime motivation for child rearing, and thus its decline is unlikely to be a major force behind a significant reduction in fertility. Second, 71 See Neher (1971) and Caldwell (1976) for earlier studies and Boldrin and Jones (2002) for a recent quan-
titative analysis.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
235
the rise in fertility rates prior to the demographic transition, in a period of improvements in credit markets, raises doubts about the significance of this mechanism. In particular, cross-section evidence does not indicate that wealthier individuals that presumably had a better access to credit markets had a smaller number of surviving offspring. On the contrary, fertility rates in the pre-demographic transition era were positively related to levels of skills, income, and wealth [e.g., Clark and Hamilton (2003)]. Exogenous shocks – Luck Becker, Murphy and Tamura (1990) advance a theory that emphasizes the role of a major exogenous shock in triggering the demographic transition, underlying the role of luck in the determination of the relative timing of the demographic transition and thus the wealth of nations.72 This theory generates predictions that are inconsistent with the observed demographic patterns in the process of development.73 Existing evidence shows that the process of industrialization and the associated increase in income per capita was accompanied by sharp increase in population growth, prior to their decline in the course of the demographic transition. In their theory, in contrast, a major shock shifts the economy from the basin of attraction of a high-fertility to a low-fertility steady-state equilibrium, generating counter-factually, a monotonic decline in fertility rates along with a monotonic rise in income per capita.
4. Unified growth theory The inconsistency of exogenous and endogenous growth models with the process of development over most of human history, induced growth theorists to develop a unified theory of economic growth that would capture in a single framework the epoch of Malthusian stagnation, the contemporary era of modern economic growth, and the underlying driving forces that triggered the recent transition between these regimes and the associated phenomenon of the Great Divergence in income per capita across countries.74 72 As they argue on page S13, “Many attempts to explain why some countries have had the best economic
performance in the past several centuries give too little attention to accidents and good fortune.” 73 Moreover, the theory suffers from several critical deficiencies in the micro-structure. First, the source of
multiplicity of equilibria in their model is the implausible assumption that the return to education increases with the aggregate level of education in society. [Browning, Hansen and Heckman (1999), for instance, show that there is weak empirical evidence in favor of this assumption.] Second, they define erroneously the lowoutput, high population growth, steady state, as a Malthusian steady-state equilibrium. Their “Malthusian” steady state, however, has none of the features of a Malthusian equilibrium. In contrast to the historical evidence about the Malthusian era, in this equilibrium (in the absence of technological change) population growth rate is not at the reproduction level. Moreover, counter-factually population growth in their “Malthusian” steady state is higher than that in the beginning of the demographic transition. Furthermore, a small positive shock to income when the economy is in the “Malthusian” steady state initially decreases fertility in contrast to the central aspect of the Malthusian equilibrium. 74 Growth theories that capture the endogenous evolution of population, technology, and output from stagnation to sustained economic growth have been established by Galor and Weil (1999, 2000), Galor and Moav
236
O. Galor
The advancement of unified growth theory was fueled by the conviction that the understanding of the contemporary growth process would be limited and distorted unless growth theory would be based on micro-foundations that would reflect the qualitative aspects of the growth process in its entirety. In particular, the hurdles faced by less developed economies in reaching a state of sustained economic growth would remain obscure unless the origin of the transition of the currently developed economies into a state of sustained economic growth would be identified and its implications would be modified to account for the additional economic forces faced by less developed economies in an interdependent world.75 The establishment of a unified growth theory has been a great intellectual challenge, requiring major methodological innovations in the construction of dynamical systems that would capture the complexity that has characterized the evolution of economies from a Malthusian epoch to a state of sustained economic growth. In light of historical evidence that suggests that the take-off from the Malthusian epoch to a state of sustained economic growth, rapid as it may appear, was a gradual process, a unified growth theory in which economies take off gradually but swiftly from an epoch of a stable Malthusian stagnation would necessitate a gradual escape from an absorbing (stable) equilibrium, in contradiction to the concept of a stable equilibrium. Thus, it has become apparent that the observed rapid, continuous phase transition would be captured by a single dynamical system, if the set of steady-state equilibria and their stability would be altered qualitatively in the process of development. As proposed by Galor and Weil (2000), during the Malthusian epoch the dynamical system would have to be characterized by a stable Malthusian equilibrium, but ultimately due to the evolution of latent state variables [i.e., the rise in a latent demand for human capital in Galor and Weil (2000) and the evolution of the distribution of genetic characteristics in Galor and Moav (2002)], the Malthusian steady-state equilibrium would vanish endogenously leaving the arena for the gravitational forces of the emerging Modern Growth Regime, and permitting the economy to take off and to converge to a modern-growth steady-state equilibrium. The observed role of the demographic transition in the shift from the Post-Malthusian Regime to the Sustained Growth Regime, and the associated non-monotonic evolution of the relationship between income per capita and population growth, added to the complexity of the desirable dynamical system. Capturing this additional transition required an endogenous reversal in the positive Malthusian effect of income on population in the second phase of industrialization, providing the reduction in fertility the observed role in the transition to a state of sustained economic growth. (2002), Hansen and Prescott (2002), Jones (2001), Kogel and Prskawetz (2001), Hazan and Berdugo (2002), Tamura (2002), Lagerlof (2003a, 2003b, 2006), Doepke (2004), Fernandez-Villaverde (2005), as well as others. The Great Divergence and its association with the transition from stagnation to growth was explored in a unified setting by Galor and Mountford (2003). 75 Although the structure of unified growth theories is based on the experience of Europe and its offshoots, since these were the areas that completed the transition from the Malthusian regime to modern growth, these theories could be modified to account for the incomplete transition of the less developed countries, integrating the significant influence of the import of pre-existing production and health technologies on their process of development.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
237
4.1. From stagnation to growth The first unified growth theory in which the endogenous evolution of population, technology, and income per capita is consistent with the process of development in the last thousands of years was advanced by Galor and Weil (2000). The theory captures the three regimes that have characterized the process of development as well as the fundamental driving forces that generated the transition from an epoch of Malthusian stagnation to a state of sustained economic growth. The theory proposes that in early stages of development economies were in the proximity of a stable Malthusian equilibrium. Technology advanced rather slowly, and generated proportional increases in output and population. The inherent positive interaction between population and technology in this epoch, however, gradually increased the pace of technological progress, and due to the delayed adjustment of population, output per capita advanced at a miniscule rate. The slow pace of technological progress in the Malthusian epoch provided a limited scope for human capital in the production process and parents, therefore, had no incentive to reallocate resources towards human capital formation of their offspring. The Malthusian interaction between technology and population accelerated the pace of technological progress and permitted a take-off to the Post-Malthusian Regime. The expansion of resources was partially counterbalanced by the enlargement of population and the economy was characterized by rapid growth rates of income per capita and population. The acceleration in technological progress ultimately increased the demand for human capital, generating two opposing effects on population growth. On the one hand, it eased households’ budget constraints, allowing the allocation of more resources for raising children. On the other hand, it induced a reallocation of resources toward child quality. In the Post-Malthusian Regime, due to the modest demand for human capital, the first effect dominated and the rise in real income permitted households to increase the number as well the quality of their children. As investment in human capital took place, the Malthusian steady-state equilibrium vanished and the economy started to be attracted by the gravitational forces of the Modern Growth Regime. The interaction between investment in human capital and technological progress generated a virtuous circle: human capital generated faster technological progress, which in turn further raised the demand for human capital, inducing further investment in child quality, and triggering the onset of the demographic transition and the emergence of a state of sustained economic growth.76 The theory suggests that the transition from stagnation to growth is an inevitable outcome of the process of development. The inherent Malthusian interaction between the level of technology and the size of the population accelerated the pace of technological progress, and eventually raised the importance of human capital in the production
76 In less developed countries the stock of human capital determines the pace of adoption of existing tech-
nologies, whereas in developed countries it determined the pace of advancement of the technological frontier.
238
O. Galor
process. The rise in the demand for human capital in the second phase of the industrial revolution and its impact on the formation of human capital as well as on the onset of the demographic transition brought about significant technological advancements along with a reduction in fertility rates and population growth, enabling economies to convert a larger share of the fruits of factor accumulation and technological progress into growth of income per capita, and paving the way for the emergence of sustained economic growth. Variations in the timing of the transition from stagnation to growth and thus in economic performance across countries (e.g., England’s earlier industrialization in comparison to China) reflect initial differences in geographical factors and historical accidents and their manifestation in variations in institutional, demographic, and cultural factors, trade patterns, colonial status, and public policy. In particular, once a technologicallydriven demand for human capital emerged in the second phase of industrialization, the prevalence of human capital promoting institutions determined the extensiveness of human capital formation, the timing of the demographic transition, and the pace of the transition from stagnation to growth. Thus, unified growth theory provides the natural framework of analysis in which variations in the economic performance across countries and regions could be examined based on the effect of variations in educational, institutional, geographical, and cultural factors on the pace of the transition from stagnation to growth. The unified theory of Galor and Weil is calibrated by Lagerlof (2006). His analysis demonstrates that the theory quantitatively replicates the observed time paths of population, income per capita, and human capital, generating (a) the Malthusian oscillations in population and output per capita during the Malthusian epoch, (b) an endogenous take-off from Malthusian stagnation that is associated with an acceleration in technological progress and is accompanied initially by a rapid increase in population growth, and (c) a rise in the demand for human capital, followed by a demographic transition and sustained economic growth. 4.1.1. Central building blocks The theory is based upon the interaction between several building blocks: the Malthusian elements, the engines of technological progress, the origin of human capital formation, and the determinants of parental choice regarding the quantity and quality of offspring. The Malthusian elements. Individuals are subjected to subsistence consumption constraint. As long as the constraint is binding, an increase in income results in an increase in population growth. Technological progress, which brings about temporary gains in income per capita, triggers therefore in early stages of development an increase in the size of the population that offsets the gain in income per capita due to the existence of diminishing returns to labor. Growth in income per capita is ultimately generated, despite diminishing returns to labor, since technological progress outpaces the rate of population growth.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
239
The forces behind technological progress in the process of development. The size of the population stimulates technological progress in early stages of development [Boserup (1965)], whereas investment in human capital is the prime engine of technological progress in more advanced stages of development. In the Malthusian era, the technological frontier was not distant from the working environment of most individuals, and the scale of the population affected the rate of technological progress due to its effect on: (a) the supply of innovative ideas, (b) the demand for new technologies, (c) the rate of technological diffusion, (d) the division of labor, and (e) the scope for trade.77 As the distance between the knowledge of an uneducated individual and the technological frontier gets larger, however, the role of human capital becomes more significant in technological advancement [e.g., Nelson and Phelps (1966)] and individuals with high levels of human capital are more likely to advance the technological frontier. The origin of human capital formation. The introduction of new technologies is mostly skill-biased although in the long run, these technologies may be either “skill biased” or “skill saving”. The “disequilibrium” brought about by technological change raises the demand for human capital.78 Technological progress reduces the adaptability of existing human capital to the new technological environment and educated individuals have a comparative advantage in adapting to the new technological environment.79 The determination of paternal decisions regarding the quantity and quality of their offspring. Individuals choose the number of children and their quality in the face of a constraint on the total amount of time that can be devoted to child-raising and labor market activities. The rise in the demand for human capital induces parents to substitute quality for quantity of children.80 4.1.2. The basic structure of the model Consider an overlapping-generations economy in which activity extends over infinite discrete time. In every period the economy produces a single homogeneous good using 77 The positive effect of the scale of the population on technological progress in the Malthusian epoch is
supported by Boserup (1965) and recent evidence by Kremer (1993). The role of the scale of the population in the modern era is, however, controversial. The distance to the technological frontier is significantly larger and population size per-se may have an ambiguous effect on technological progress, if it comes on the account of population quality. 78 If the return to education rises with the level of technology the qualitative results would not be affected. Adopting this mechanism, however, would be equivalent to assuming that changes in technology were skillbiased throughout human history. Although on average technological change may have been skilled biased, Galor and Weil’s mechanism is consistent with periods in which the characteristics of new technologies could be defined as unskilled-biased, most notably, in the first phase of the industrial revolution. 79 Schultz (1975) cites a wide range of evidence in support of this assumption. More recently, Foster and Rosenzweig (1996) find that technological change during the green revolution in India raised the return to schooling, and that school enrollment rate responded positively to this higher return. The effect of technological transition on the return to human capital is at the center of the theoretical approach of Galor and Tsiddon (1997), Galor and Moav (2000), and Hassler and Rodriguez Mora (2000). 80 The existence of a trade-off between quantity and quality of children is supported empirically [e.g., Rosenzweig and Wolpin (1980) and Hanushek (1992)].
240
O. Galor
land and efficiency units of labor as inputs. The supply of land is exogenous and fixed over time whereas the number of efficiency units of labor is determined by households’ decisions in the preceding period regarding the number and level of human capital of their children. Production of final output Production occurs according to a constant returns to scale technology that is subject to endogenous technological progress. The output produced at time t, Yt , is Yt = Htα (At X)1−α ,
(1)
where Ht is the aggregate quantity of efficiency units of labor employed in period t, X is land employed in production in every period t, At represents the endogenously determined technological level in period t, and At X are therefore the “effective resources” employed in production in period t, and α ∈ (0, 1). Output per worker produced at time t, yt , is yt = hαt xt1−α ,
(2)
where ht ≡ Ht /Lt is the level of efficiency units of labor per worker, and xt ≡ (At X)/Lt is the level of effective resources per worker at time t. Suppose that there are no property rights over land.81 The return to land is therefore zero, and the wage per efficiency unit of labor is equal to the output per efficiency unit of labor: wt = (xt / ht )1−α .
(3)
Preferences and budget constraints In each period t, a generation that consists of Lt identical individuals joins the labor force. Each individual has a single parent. Members of generation t (those who join the labor force in period t) live for two periods. In the first period of life (childhood), t − 1, individuals consume a fraction of their parental unit time endowment. The required time increases with children’s quality. In the second period of life (parenthood), t, individuals are endowed with one unit of time, which they allocate between child rearing and labor force participation. They choose the optimal mixture of quantity and quality of (surviving) children and supply their remaining time in the labor market, consuming their wages. Individuals’ preferences are represented by a utility function defined over consumption above a subsistence level c˜ > 0, as well as over the quantity and quality (measured 81 The modeling of the production side is based upon two simplifying assumptions. First, capital is not
an input in the production function, and second the return to land is zero. Alternatively it could have been assumed that the economy is small and open to a world capital market in which the interest rate is constant. In this case, the quantity of capital will be set to equalize its marginal product to the interest rate, while the price of land will follow a path such that the total return on land (rent plus net price appreciation) is also equal to the interest rate. Allowing for capital accumulation and property rights over land would complicate the model to the point of intractability, but would not affect the qualitative results.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
241
by human capital) of their (surviving) children:82 ut = (ct )1−γ (nt ht+1 )γ ,
γ ∈ (0, 1),
(4)
where ct is the consumption of individual of generation t, nt is the number of children of individual t, and ht+1 is the level of human capital of each child.83 The utility function is strictly monotonically increasing and strictly quasi-concave, satisfying the conventional boundary conditions that assure, for a sufficiently high income, the existence of an interior solution for the utility maximization problem. However, for a sufficiently low level of income the subsistence consumption constraint is binding and there is a corner solution with respect to the consumption level.84 Individuals choose the number of children and their quality in the face of a constraint on the total amount of time that can be devoted to child-raising and labor market activities. For simplicity, only time is required in order to produce child quantity and quality.85 Let τ + et+1 be the time cost for a member i of generation t of raising a child with a level of education (quality) et+1 . That is, τ is the fraction of the individual’s unit time endowment that is required in order to raise a child, regardless of quality, and et+1 is the fraction of the individual’s unit time endowment that is devoted for the education of each child.86 Consider members of generation t who are endowed with ht efficiency units of labor at time t. Define potential income, zt , as the earning if the entire time endowment is devoted to labor force participation, earning the competitive market wage, wt , per efficiency unit. The potential income, zt ≡ wt ht , is divided between consumption, ct , and expenditure on child rearing (quantity as well as quality), evaluated according to the value of the time cost, i.e., wt ht [τ + et+1 ], per child. Hence, in the second period of life (parenthood), the individual faces the budget constraint wt ht nt (τ + et+1 ) + ct wt ht ≡ zt .
(5)
82 For simplicity parents derive utility from the expected number of surviving offspring and the parental cost
of child rearing is associated only with surviving children. A more realistic cost structure would not affect the qualitative features of the theory. 83 Alternatively, the utility function could have been defined over consumption above subsistence rather than over a consumption set that is truncated from below by the subsistence consumption constraint. In particular, if ut = (ct − c) ˜ (1−γ ) (nt ht+1 )γ , the qualitative analysis would not be affected, but the complexity of the dynamical system would be greatly enhanced. The income expansion path would be smooth, transforming continuously from being nearly vertical for low levels of potential income to asymptotically horizontal for high levels of potential income. The subsistence consumption constraint would therefore generate the Malthusian effect of income on population growth at low income levels. 84 The subsistence consumption constraint generates the positive income elasticity of population growth at low income levels, since higher income allows individuals to afford more children. 85 If both time and goods are required in order to produce child quality, the process we describe would be intensified. As the economy develops and wages increase, the relative cost of a quality child will diminish and individuals will substitute quality for quantity of children. 86 τ is assumed to be sufficiently small so as to assure that population can have a positive growth rate. That is, τ < γ .
242
O. Galor
The production of human capital Individuals’ level of human capital is determined by their quality (education) as well as by the technological environment. Technological progress reduces the adaptability of existing human capital for the new technological environment (the ‘erosion effect’). Education, however, lessens the adverse effects of technological progress. That is, skilled individuals have a comparative advantage in adapting to the new technological environment. In particular, the time required for learning the new technology diminishes with the level of education and increases with the rate of technological change. The level of human capital of children of a member i of generation t, hit+1 , is an i , increasing strictly concave function of their parental time investment in education, et+1 and a decreasing strictly convex function of the rate of technological progress, gt+1 : ht+1 = h(et+1 , gt+1 ),
(6)
where gt+1 ≡ (At+1 − At )/At . Education lessens the adverse effect of technological progress. That is, technology complements skills in the production of human capital i ,g (i.e., heg (et+1 t+1 ) > 0). In the absence of investment in quality, each individual has a basic level human capital that is normalized to 1 in a stationary technological environment, i.e., h(0, 0) = 1.87 Optimization Members of generation t choose the number and quality of their children, and therefore their own consumption, so as to maximize their intertemporal utility function subject to the subsistence consumption constraint. Substituting (5)–(6) into (4), the optimization problem of a member of generation t is: γ 1−γ {nt , et+1 } = argmax wt ht 1 − nt (τ + et+1 ) (7) nt h(et+1 , gt+1 subject to: ˜ wt ht 1 − nt (τ + et+1 ) c; (nt , et+1 ) 0. Hence, as long as potential income at time t is sufficiently high so as to assure that ct > c˜ (i.e., as long as zt ≡ wt ht , is above the level of potential income at which ˜ − γ ))), the fraction of the subsistence constraint is just binding, (i.e., zt > z˜ ≡ c/(1 time spent by individual t raising children is γ , while 1 − γ is devoted for labor force participation. However, if zt z˜ , the subsistence constraint is binding, the fraction of time necessary to assure subsistence consumption, c, ˜ is larger than 1−γ and the fraction 87 For simplicity, investment in quality is not beneficial in a stationary technological environment, i.e.,
he (0, 0) = 0, and in the absence of investment in education, there exists a sufficiently rapid technological progress, that due to the erosion effect renders the existing human capital obsolete (i.e., limg→∞ h(0, gt+1 ) = 0). Furthermore, although the potential number of efficiency units of labor is diminished due to the transition from the existing technological state to a superior one (due to the erosion effect), each individual operates with a superior level of technology and the productivity effect is assumed to dominate. That is, ∂yt /∂gt > 0.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
243
Figure 37. Preferences, constraints, and income expansion path.
of time devoted for child rearing is therefore below γ . That is, γ if zt z˜ , nt [τ + et+1 ] = 1 − [c/w ˜ t ht ] if zt z˜ .
(8)
Figure 37 shows the effect of an increase in potential income zt on the individual’s allocation of time between child rearing and consumption. The income expansion path is vertical as long as the subsistence consumption constraint is binding. As the wage per efficiency unit of labor increases in this income range, the individual can generate the subsistence consumption with a smaller labor force participation and the fraction of time devoted to child rearing increases. Once the level of income is sufficiently high such that the subsistence consumption constraint is not binding, the income expansion path becomes horizontal at a level γ in terms of time devoted to child rearing. Furthermore, the optimization with respect to et+1 implies that the level of education chosen by members of generation t for their children, et+1 , is an increasing function of gt+1 . ˆ = 0 if gt+1 g, et+1 = e(gt+1 ) (9) > 0 if gt+1 > gˆ where e (gt+1 ) > 0 and e (gt+1 ) < 0 ∀gt+1 > gˆ > 0.88 Hence, regardless of whether potential income is above or below z˜ , increases in wages will not change the division of child-rearing time between quality and quantity. However, the division between time 88 e (g t+1 ) depends upon the third derivatives of the production function of human capital. e (gt+1 ) is
assumed to be concave, which appears plausible.
244
O. Galor
spent on quality and time spent on quantity is affected by the rate of technological progress, which changes the return to education. Furthermore, substituting (9) into (8), it follows that nt is: γ b if zt z˜ , τ + e(gt+1 ) ≡ n (gt+1 ) nt = (10) 1 − [c/z ˜ t] ≡ na gt+1 , z(et, gt , xt ) if zt z˜ τ + e(gt+1 ) where zt ≡ wt ht = z(et, gt , xt ) as follows from (3) and (6). Hence, as follows from the properties of e(gt+1 ), nb (gt+1 ), and na (gt+1 , zt ): (a) An increase in the rate of technological progress reduces the number of children and increases their quality, i.e., ∂nt /∂gt+1 0 and ∂et+1 /∂gt+1 0. (b) If the subsistence consumption constraint is binding (i.e., if parental potential income is below z˜ ), an increase in parental potential income raises the number of children, but has no effect on their quality, i.e., ∂nt /∂zt > 0 and ∂et+1 /∂zt = 0 if zt < z˜ . (c) If the subsistence consumption constraint is not binding (i.e., if parental potential income is above z˜ ), an increase in parental potential income does not affect the number of children and their quality, i.e., ∂nt /∂zt = ∂et+1 /∂zt = 0
if zt > z˜ .
Technological progress Suppose that technological progress, gt+1 , that takes place between periods t and t + 1 depends upon the education per capita among the working generation in period t, et , and the population size in period t, Lt :89 At+1 − At = g(et , Lt ), (11) At where for et 0 and a sufficiently large population size Lt , g(0, Lt ) > 0, gi (et , Lt ) > 0, and gii (et , Lt ) < 0, i = et , Lt .90 Hence, for a sufficiently large population size, gt+1 ≡
89 While the role of the scale effect in the Malthusian epoch is essential, none of the existing results depend
on the presence or the absence of the scale effect in the modern era. The functional form of technological progress given in (11) can capture both the presence and the absence of the scale effect in the modern era. In particular, the scale effect can be removed, once investment in education is positive, assuming for instance that limL→∞ gL (et , L) = 0 for et > 0. 90 For a sufficiently small population the rate of technological progress is strictly positive only every several periods. Furthermore, the number of periods that pass between two episodes of technological improvement declines with the size of population. These assumptions assure that in early stages of development the economy is in a Malthusian steady state with zero growth rate of output per capita, but ultimately the growth rates is positive and slow. If technological progress would occur in every time period at a pace that increases with the size of population, the growth rate of output per capita would always be positive, despite the adjustment in the size of population.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
245
the rate of technological progress between time t and t + 1 is a positive, increasing, strictly concave function of the size of adult population and the level of education of the working generation at time t. Furthermore, the rate of technological progress is positive even if labor quality is zero. The state of technology at time t + 1, At+1 , is therefore At+1 = (1 + gt+1 )At ,
(12)
where the state of technology at time 0 is given at a level A0 . Population The size of the adult population at time t + 1, Lt+1 , is Lt+1 = nt Lt ,
(13)
where Lt is the size the adult of population at time t and nt is the number of children per person; L0 is given. Hence, given (10), the evolution of the adult population over time is b if zt z˜ , n (g t+1 )Lt Lt+1 = (14) na gt+1 , z(et, gt , xt ) Lt if zt z˜ . Effective resources The evolution of effective resources per worker, xt ≡ (At X)/Lt , is determined by the evolution of population and technology. The level of effective resources per worker in period t + 1 is 1 + gt+1 (15) xt , nt where x0 ≡ A0 X/L0 is given. Furthermore, as follows from (10) and (11) [1 + g(et , Lt )][τ + e(g(et , Lt ))] xt ≡ φ b (et , Lt )xt if zt z˜ , γ xt+1 = [1 + g(et , Lt )][τ + e(g(et , Lt ))] xt ≡ φ a (et , gt , xt , Lt )xt if zt z˜ , 1 − [c/z(e ˜ t , gt , xt )] (16) where φeb (et , Lt ) > 0 and φxa (et , gt , xt , Lt ) < 0, ∀et 0. xt+1 =
4.1.3. The dynamical system The development of the economy is fully determined by a sequence {et , gt , xt , Lt }∞ t=0 that satisfies (9), (11), (14), and (16), in every period t and describes the joint evolution of education, technological progress, effective resources per capita, and population over time. The dynamical system is characterized by two regimes. In the first regime the subsistence consumption constraint is binding and the evolution of the economy is governed by a four dimensional non-linear first-order autonomous system: x a t+1 = φ (et , gt , xt , Lt )xt , e t+1 = e g(et , Lt ) , (17) for zt z˜ g t , Lt ), t+1 = g(e a Lt+1 = n g(et , Lt ), z(et, gt , xt ) Lt
246
O. Galor
Figure 38. The evolution of technology, gt , education, et , and effective resources, xt : small population.
where the initial conditions e0 , g0 , x0 , and L0 are historically given. In the second regime the subsistence consumption constraint is not binding and the evolution of the economy is governed by a three-dimensional system: xt+1 = φb (et , L)xt , et+1 = e g(et , L) , (18) for zt z˜ . b Lt+1 = n g(et , Lt ) Lt In both regimes, however, the analysis of the dynamical system is greatly simplified by the fact that the evolution of et and gt is independent of whether the subsistence constraint is binding, and by the fact that, for a given population size L, the joint evolution of et and gt is determined independently of xt . The education level of workers in period t + 1 depends only on the level of technological progress expected between period t and period t + 1, while given L, technological progress between periods t and t + 1 depends only on the level of education of workers in period t. Thus, for a given population size L, the dynamics of technology and education can be analyzed independently of the evolution of resources per capita. A. The dynamics of technology and education The evolution of technology and education, for a given population size L, is characterized by the sequence {gt , et ; L}∞ t=0 that satisfies in every period t the equations gt+1 = g(et ; L), and et+1 = e(gt+1 ). Although this dynamical sub-system consists of two independent one-dimensional, nonlinear first-order difference equations, it is more revealing to analyze them jointly. In light of the properties of the functions e(gt+1 ) and g(et ; L) this dynamical subsystem is characterized by three qualitatively different configurations, which are depicted in Figures 38A, 39A and 40A. The economy shifts endogenously from one configuration to another as population increases and the curve g(et ; L) shifts upward to account for the effect of an increase in population.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
247
Figure 39. The evolution of technology, gt , education, et , and effective resources, xt : moderate population.
Figure 40. The evolution of technology, gt , education, et , and effective resources, xt : large population.
In Figure 38A, for a range of small population size, the dynamical system is characterized by globally stable steady-state equilibria, (e(L), ¯ g(L)) ¯ = (0, g l (L)), where g l (L) increases with the size of the population while the level of education remains unchanged. In Figure 39A, for a range of moderate population size, the dynamical system is characterized by three steady-state equilibria, two locally stable steady-state equi¯ g(L)) ¯ = (eh (L), g h (L)), and an interior libria: (e(L), ¯ g(L)) ¯ = (0, g l (L)) and (e(L), u u unstable steady state (e(L), ¯ g(L)) ¯ = (e (L), g (L)), where (eh (L), g h (L)) and g l (L) increase monotonically with the size of the population. Finally, in Figure 40A, for a range of large population sizes, the dynamical system is characterized by globally stable steady-state equilibria, (e(L), ¯ g(L)) ¯ = (eh (L), g h (L)), where eh (L) and g h (L) increase monotonically with the size of the population.
248
O. Galor
B. Global dynamics This section analyzes the evolution of the economy from the Malthusian Regime, through the Post-Malthusian Regime, to the demographic transition and the Modern Growth Regime. The global analysis is based on a sequence of phase diagrams that describe the evolution of the system, within each regime, for a given population size, and the transition between these regimes as population increases in the process of development. Each of the phase diagrams is a two-dimensional projection in the plain (et , xt ; L), of the three-dimensional system in the space {et , gt , xt ; L}. The phase diagrams, depicted in Figures 38B, 39B, and 40B contain three elements: the Malthusian Frontier, which separates the regions in which the subsistence constraint is binding from those where it is not; the XX locus, which denotes the set of all triplets (et , gt , xt ; L) for which effective resources per worker are constant; and the EE locus, which denotes the set of all pairs (et , gt ; L) for which the level of education per worker is constant. The Malthusian Frontier As was established in (17) and (18) the economy exits from the subsistence consumption regime when potential income, zt , exceeds the critical level z˜ . This switch of regime changes the dimensionality of the dynamical system from four to three. Let the Malthusian Frontier be the set of all triplets of (et , xt , gt ; L) for which individuals’ income equal z˜ .91 Using the definitions of zt and z˜ , it follows from (2) and (6) ˜ )}. that the The Malthusian Frontier, MM ≡ {(et , xt , gt ; L): xt1−α h(et , gt )α = c/(1−γ Let the Conditional Malthusian Frontier be the set of all pairs (et , xt ; L) for which, conditional on a given technological level gt , individuals incomes equal z˜ . Following the definitions of zt and z˜ , Equations (2) and (6) imply that the Conditional Malthusian (1−α) Frontier, MM|gt , is MM|gt ≡ {(et , xt ; L): xt h(et , gt )α = c/(1 ˜ − γ ) | gt }, where xt is a decreasing strictly convex function of et along the MM|gt locus. Hence, the Conditional Malthusian Frontier, as depicted in Figures 38B–40B, is a strictly convex, downward sloping, curve in the (et , xt ) space. Furthermore, it intersects the xt axis and approaches asymptotically the et axis as xt approaches infinity. The frontier shifts upward as gt increases in the process of development. The XX locus Let XX be the locus of all triplets (et , gt , xt ; L) such that the effective resources per worker, xt , is in a steady state: XX ≡ {(et , xt , gt ; L): xt+1 = xt }. As follows from (15), along the XX locus the growth rates of population and technology are equal. Above the Malthusian frontier, the fraction of time devoted to childrearing is independent of the level of effective resources per worker. In this case, the growth rate of population will just be a negative function of the growth rate of technology, since for higher technology growth, parents will spend more of their resources 91 Below the Malthusian Frontier, the effect of income on fertility will be positive, while above the frontier
there will be no effect of income on fertility. Thus, the Malthusian Frontier separates the Malthusian and Post-Malthusian Regimes, on the one hand, from the Modern Growth regime, on the other, and crossing this frontier is associated with the demographic transition.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
249
on child quality and thus less on child quantity. Thus there will be a particular level of technological progress which induces an equal rate of population growth. Since the growth rate of technology is, in turn, a positive function of the level of education, this rate of technology growth will correspond to a particular level of education, denoted e. ˆ Below the Malthusian Frontier, the growth rate of population depends on the level of effective resources per capita, x, as well as on the growth rate of technology. The lower is x, the smaller the fraction of the time endowment devoted to child-rearing, and so the lower is population growth. Thus, below the Malthusian frontier, a lower value of effective resources per capita would imply that lower values of technology growth (and thus education) would be consistent with population growth being equal to technology growth. Thus, as drawn in Figures 38B, 39B, and 40B, lower values of x are associated with lower values of e on the part of the XX locus that is below the Malthusian frontier. If the subsistence consumption constraint is not binding, it follows from (16) that for ˆ < eh (L), such that xt ∈ XX.92 zt z˜ , there exists a unique value 0 < e(L) xt+1 − xt
> 0 if et > e(L), ˆ ˆ = 0 if et = e(L), ˆ < 0 if et < e(L).
(19)
Hence, the XX Locus, as depicted in Figures 38B, 39B, and 40B is a vertical line above the Conditional Malthusian Frontier at a level e(L). ˆ If the subsistence constraint is binding, the evolution of xt is based upon the rate of technological change, gt , the effective resources per-worker, xt as well as the quality of the labor force, et . Let XX |gt be the locus of all pairs (et , xt ; L) such that xt+1 = xt , for a given level of gt . That is, XX |gt ≡ {(et , xt ; L): xt+1 = xt | gt }. It follows from (16) ˆ there exists a single-valued function xt = x(et ) that for zt z˜ , and for 0 et e(L), such that (x(et ), et ) ∈ XX |gt . < 0 xt+1 − xt = 0 >0
ˆ if (et , xt ) > et , x(et ) for 0 et e(L), for 0 e e(L), ˆ if xt = x(et ) t ˆ , or et > e(L) ˆ . if (et , xt ) < et , x(et ) for 0 et e(L) (20)
Hence, without loss of generality, the locus XX |gt is depicted in Figure 38 as an upˆ XX |gt is strictly below ward slopping curve in the space (et , xt ), defined for et e(L). ˆ and the two coincides at the Conditional Malthusian Frontier for value of et < e(L), e(L). ˆ Moreover, the Conditional Malthusian Frontier, the XX locus, and the XX |gt locus, coincide at (e(L), ˆ x(L)). ˆ
92 In order to simplify the exposition without affecting the qualitative nature of the dynamical system, the
parameters of the model are restricted so as to assure that the XX locus is non-empty when zt z˜ . That is, gˆ < (γ /τ ) − 1 < g(eh (L0 ), L0 ).
250
O. Galor
The EE locus Let EE be the locus of all triplets (et , gt , xt ; L) such that the quality of labor, et , is in a steady state: EE ≡ {(et , xt , gt ; L): et+1 = et }. As follows from (9) and (11), et+1 = e(g(et ; L)) and thus, for a given population size, the steady-state values of et are independent of the values of xt and gt . The locus EE evolves through three phases in the process of development, corresponding to the three phases that describe the evolution of education and technology, as depicted in Figures 38A, 39A, and 40A. In early stages of development, when population size is sufficiently small, the joint evolution of education and technology is characterized by a globally stable temporary steady-state equilibrium, (e(L), ¯ g(L)) ¯ = (0, g l (L)), as depicted in Figure 38A. The corresponding EE locus, depicted in the space (et , xt ; L) in Figure 38B, is vertical at the level e = 0, for a range of small population sizes. Furthermore, for this range, the global dynamics of et are given by: = 0 if et = 0, et+1 − et (21) < 0 if et > 0. In later stages of development as population size increases sufficiently, the joint evolution of education and technology is characterized by multiple locally stable temporary steady-state equilibria, as depicted in Figure 39A. The corresponding EE locus, depicted in the space (et , xt ; L) in Figure 39B, consists of three vertical lines corresponding to the three steady-state equilibria for the value of et . That is, e = 0, e = eu (L), and e = eh (L). The vertical line e = eu (L) shifts leftward, and e = eh (L) shifts rightward as population size increases. Furthermore, the global dynamics of et in this configuration are given by: h < 0 if 0 < et < eu (L) or et > e (L), et+1 − et = 0 if et = 0, eu (L), eh (L) , (22) u h > 0 if e (L) < et < e (L). In mature stages of development when population size is sufficiently large, the joint evolution of education and technology is characterized by a globally stable steady-state equilibrium, (e(L), ¯ g(L)) ¯ = (eh (L), g h (L)), as depicted in Figure 40A. The corresponding EE locus, as depicted in Figure 40B in the space (et , xt ; L), is vertical at the level e = eh (L). This vertical line shifts rightward as population size increases. Furthermore, the global dynamics of et in this configuration are given by: > 0 if 0 et < eh (L), et+1 − et = 0 if et = eh (L), (23) < 0 if et > eh (L). Conditional steady-state equilibria In early stages of development, when population size is sufficiently small, the dynamical system, as depicted in Figure 38B, is characterized by a unique and globally stable conditional steady-state equilibrium.93 It is 93 Since the dynamical system is discrete, the trajectories implied by the phase diagrams do not necessarily
approximate the actual dynamic path, unless the state variables evolve monotonically over time. As shown,
Ch. 4: From Stagnation to Growth: Unified Growth Theory
251
given by a point of intersection between the EE locus and the xt+1 = xt locus. That is, ¯ is conditional on a given technological level, gt , the Malthusian steady state (0, x(L)) globally stable.94 In later stages of development as population size increases sufficiently, the dynamical system as depicted in Figure 39B is characterized by two conditional steady-state equilibria. The Malthusian conditional steady-state equilibrium is locally stable, whereas the steady-state equilibrium (eu (L), x u (L)) is a saddle point.95 For education levels above eu (L) the system converges to a stationary level of education eh (L) and possibly to a steady-state growth rate of xt . In mature stages of development when population size is sufficiently large, the system convergences globally to an educational level eh (L) and possibly to a steady-state growth rate of xt . 4.1.4. From Malthusian stagnation to sustained growth The economy evolves from an epoch of Malthusian stagnation through the PostMalthusian Regime to the demographic transition and a Modern Growth Regime. This pattern and the prime driving forces in this transition emerge from the phase diagrams depicted in Figures 38–40. Consider an economy in early stages of development. Population size is relatively small and the implied slow rate of technological progress does not provide an incentive to invest in the education of children. As depicted in Figure 38A, the interaction between education, et , and the rate of technological change, gt , for a constant small population, L, is characterized by a globally stable steady-state equilibrium (0, g l (L)), where education is zero and the rate of technological progress is slow. This steady-state equilibrium corresponds to a globally stable conditional Malthusian steady-state equilibrium, depicted in Figure 38B. For a constant small population, L, and for a given rate of technological progress, effective resources per capita, as well as the level of education are constant, and output per capita is therefore constant as well. Moreover, shocks to population or resources will be resolved in a Malthusian fashion. As population grows slowly in reaction to technological progress, the g(et+1 , L) locus, depicted in Figure 38A, gradually shifts upward. The steady-state equilibrium shifts vertically upward reflecting small increments in the rate of technological progress, while the level of education remains constant at a zero level. Similarly, the conditional Malthusian steady-state equilibrium drawn in Figure 38B shifts vertically upward, as the XX the evolution of et is monotonic, whereas the evolution and convergence of xt may be oscillatory. Nonmonotonicity in the evolution of xt may arise only if e < eˆ and it does not affect the qualitative description of the system. Furthermore, if φxa (et , gt , xt )xt > −1 the conditional dynamical system is locally non-oscillatory. The phase diagrams in Figures 38A–40A are drawn under the assumptions that assure that there are no oscillations. 94 The local stability of the steady-state equilibrium (0, x(g )) can be derived formally. The eigenvalues of t the Jacobian matrix of the conditional dynamical system evaluated at the conditional steady-state equilibrium are both smaller than one (in absolute value). 95 Convergence to the saddle point takes place only if the level of education is eu . That is, the saddle path is the entire vertical line that corresponds to et = eu .
252
O. Galor
locus shifts upward. However, output per capita remains initially constant at the subsistence level and eventually creeps forward at a miniscule rate. Over time, the slow growth in population that takes place in the Malthusian Regime raises the rate of technological progress and shifts the g(et+1 , L) locus in Figure 38A sufficiently upward, generating a qualitative change in the dynamical system depicted in Figure 39A. The dynamical system of education and technology, for a moderate population, is characterized by multiple, history-dependent, stable steady-state equilibria: the steady-state equilibria (0, g l (L)) and (eh (L), g h (L)) are locally stable, whereas (eu (L), g u (L)) is unstable. Given the initial conditions, in the absence of large shocks, the economy remains in the vicinity of the low steady-state equilibrium (0, g l (L)), where education is still zero but the rate of technological progress is moderate. These steady-state equilibria correspond to a multiple locally stable conditional Malthusian steady-state equilibrium, depicted in Figure 39B: Malthusian steady state, characterized by constant resources per capita, slow technological progress, and no education, and a modern-growth steady state, characterized by a high level of education, rapid technological progress, growing income per capita, and moderate population growth. However, since the economy starts in the vicinity of Malthusian steady state, it remains there.96 As the rate of technological progress continues to rise in reaction to the increasing population size, the g(et+1 , Lt ) locus shifts upward further and ultimately, as depicted in Figure 40, the dynamical system experiences another qualitative change. The Malthusian steady-state equilibrium vanishes, and the economy is characterized by a unique globally stable modern steady-state equilibrium (eh (L), g h (L)) characterized by high levels of education and technological progress. Increases in the rate of technological progress and the level of education feed back on each other until the economy converges rapidly to the stable modern steady-state equilibrium. The increase in the pace of technological progress has two opposing effects on the evolution of population. On the one hand, it eases households’ budget constraints, allowing the allocation of more resources for raising children. On the other hand, it induces a reallocation of these additional resources toward child quality. In the Post-Malthusian Regime, due to the limited demand for human capital, the first effect dominates and the rise in real income permits households to increase their family size as well the quality of each child. The interaction between investment in human capital and technological progress generates a virtuous circle: human capital formation prompts faster technological progress, further raising the demand for human capital, inducing further investment in child quality, and eventually, as the economy crosses the Malthusian frontier, triggering a demographic transition. The offsetting effect of population
96 Large shocks to education or technological progress would permit the economy to jump to the modern-
growth steady state, but this possibility appears inconsistent with the evidence.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
253
growth on the growth rate of income per capita is eliminated and the interaction between human capital accumulation and technological progress permits a transition to a state of sustained economic growth. In the Modern Growth Regime, resources per capita rise as technological progress outstrips population growth. Provided that population size is constant (i.e., population growth is zero), the levels of education and technological progress and the growth rates of resources per capita and thus output per capita are constant in the modern-growth steady-state equilibrium.97 4.1.5. Major hypotheses and their empirical assessment The theory generates several hypotheses about the evolution of population, human capital and income per capita in the process of development, underlying the roles of the inherent interaction between population and technology in the Malthusian epoch, as well as the formation of human capital in the second phase of industrialization and the associated demographic transition, in the emergence of a state of sustained economic growth. Main hypotheses: (H1) During the initial phases of the Malthusian epoch the growth rate of output per capita is nearly zero and the growth rate of population is miniscule, reflecting the sluggish pace of technological progress and the full adjustment of population to the expansion of resources. In the later phases of the Malthusian epoch, the increasing rate of technological progress, along with the inherent delay in the adjustment of population to the rise in income per capita, generated positive but very small growth rates of output per capita and population. The hypothesis is consistent with the evidence, provided in Section 2.1, about the evolution of the world economy in the Malthusian epoch. In particular, the infinitesimal pace of resource expansion in the first millennium was reflected in a miniscule increase of the Western European population (from 24.7 million people in 1 AD to 25.4 million in 1000 AD), along with a zero average growth rate of output per capita. The more rapid, but still very slow expansion of resources in the period 1000–1500, permitted the Western European population to grow at a slow average rate of 0.16% per year (from 25 million
97 If population growth is positive in the Modern Growth Regime, then education and technological progress
continue to rise. Similarly, if population growth is negative, they fall. In fact, the model makes no firm prediction about what the growth rate of population will be in the Modern Growth Regime, other than that population growth will fall once the economy exits from the Malthusian region. If the growth rate of technology is related to the growth rate of population, rather than to its level, then there exists a steady state characterized by modern-growth in which the growth rates of population and technology would be constant. Further, such a steady state would be stable: if population growth fell, the rate of technological progress would also fall, inducing a rise in fertility.
254
O. Galor
in 1000 to 57 million in 1500), along with a slow average growth rate of income per capita at a rate of about 0.13% per year. Resource expansion over the period 1500–1820 had a more significant impact on the Western European population that grew at an average pace of 0.26% per year (from 57 million in 1500 to 133 million in 1820), along with a slightly faster average growth rate of income per capita at a rate of about 0.15% per year. (H2) The reinforcing interaction between population and technology during the Malthusian epoch increased the size of the population sufficiently so as to support a faster pace of technological progress, generating the transition to the Post-Malthusian Regime. The growth rates of output per capita increased significantly, but the positive Malthusian effect of income per capita on population growth was still maintained, generating a sizable increase in population growth, and offsetting some of the potential gains in income per capita. Moreover, human capital accumulation did not play a significant role in the transition to the Post-Malthusian Regime and thus in the early take-off in the first phase of the Industrial Revolution. The hypothesis is consistent with the evidence, provided in Section 2.2, about the evolution of the world economy in the Post-Malthusian Regime. In particular, the acceleration in the pace of resource expansion in the period 1820–1870, increased the Western European population from 133 million people in 1820 to 188 million in 1870, while the average growth rate of output per capita over this period increased significantly to 0.95% per year. Furthermore, historical evidence suggests that the industrial demand for human capital increased only in the second phase of the Industrial Revolution. As shown by Clark (2003), human capital formation prior to the Industrial Revolution, as well as in its first phase, occurred in an era in which the market rewards for skill acquisition were at historically low levels.98 (H3) The acceleration in the rate of technological progress increased the industrial demand for human capital in the later part of Post-Malthusian Regime (i.e., the second phase of industrialization), inducing significant investment in human capital, and triggering the demographic transition and a rapid pace of economic growth. This hypothesis is consistent with the evidence, provided in Section 2.3, and partly depicted in Figure 41, about the significant rise in the industrial demand for human capital in the second phase of the Industrial Revolution, the marked increase in educational attainment, and the decline in fertility rates, that occurred in association with the acceleration in the growth rate of output per capita. In particular, the predicted timing of the acceleration in the growth rate of output per capita, is consistent with the revisionist
98 The rise in human capital formation over this period may reflect religious, cultural and social forces, as well
as the rise in valuation for offspring quality due to the forces of natural selection, as discussed in Section 5.2.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
255
Figure 41. The sharp rise in the growth rate of real GDP per capita and its association with investment in education and fertility decline: England 1485–1920. Sources: Clark (2001), Feinstein (1972), Flora, Kraus and Pfenning (1983), Wrigley and Schofield (1981).
view on the British Industrial Revolution [e.g., Crafts and Harley (1992), Clark (2001), and Voth (2003)] that suggests that the first phase of the Industrial Revolution in England was characterized by a moderate increase in the growth rate of output per capita, whereas the “take-off”, as depicted in Figure 41, occurred only in the 1860s. Furthermore, quantitative analysis of unified growth theories by Doepke (2004), Fernandez-Villaverde (2005), Lagerlof (2006), and Pereira (2003) indeed suggest that the rise in the demand for human capital was a significant force behind the demographic transition and the emergence of a state of sustained economic growth.99 Moreover, the theory is consistent with the observed simultaneous onset of the demographic transition across Western European countries that differed significantly in their income per capita. It suggests that a technologically-driven universal rise in the demand for human capital in Western Europe (as documented in Section 2.3.3) generated this simultaneous transition. It should be noted that the lack of clear evidence about the increase in the return to human capital in the second phase of the Industrial Revolution does not indicate the absence of a significant increase in the demand for human capital over this period. The sizable increase in schooling that took place in the 19th century and in particular the introduction of public education that lowered the cost of educa-
99 The rise in the demand for human capital in Fernandez-Villaverde (2005) is based on capital-skill comple-
mentarity, and is indistinguishable from the complementarity between technology and skills (in the short run) that is maintained by Galor and Weil (2000).
256
O. Galor
tion (e.g., The Education Act of 1870), generated significant increase in the supply of educated workers that may have prevented a rise in the return to education.100 (H4) The growth process is characterized by stages of development and it evolves non-linearly. Technological leaders experienced a monotonic increase in the growth rates of their income per capita. Their growth was rather slow in early stages of development, it increased rapidly during the take-off from the Malthusian epoch, and it continued to rise, often stabilizing at higher levels. In contrast, technological followers that made the transition to sustained economic growth, experienced a non-monotonic increase in the growth rates of their income per capita. Their growth rates was rather slow in early stages of development, it increased rapidly in the early stages of the take-off from the Malthusian epoch, boosted by the adoption of technologies from the existing technological frontier. However, once these economies reached the technological frontier, their growth rates dropped to the level of the technological leaders. (H5) The differential timing of the take-off from stagnation to growth across economies generated convergence clubs characterized by a group of poor countries in the vicinity of the Malthusian equilibrium, a group of rich countries in the vicinity of the sustained growth equilibrium, and a third group in the transition from one club to another.101 These hypotheses are consistent with Maddison’s (2001) evidence about the growth process in the last 250 years, as well as with contemporary cross section evidence. These studies suggest that the growth process is characterized by multiple growth regimes [e.g., Durlauf and Johnson (1995)] and thus with non-linearities in the evolution of growth rates [e.g., Durlauf and Quah (1999), Bloom, Canning and Sevilla (2003), and Graham and Temple (2004)]. Moreover, this research demonstrates that the evolution of income per-capita across countries is characterized by divergence in the past two centuries along with a tendency towards the emergence of a twin peak distribution [Quah (1996, 1997), Jones (1997), and Pritchett (1997)].102 4.2. Complementary theories Subsequent theories of economic growth in the very long run demonstrate that the unified theory of economic growth can be augmented and fortified by additional characteristics of the transition from stagnation to growth without altering the fundamental 100 Some of this supply response was a direct reaction of the potential increase in the return to human capital,
and thus may only operate to partially offset the increase in the return to human capital, but the reduction in the cost of education via public schooling, generated an additional force that operated towards a reduction in the return to human capital. 101 For the definition of club convergence see Azariadis (1996) and Galor (1996). 102 Other studies that focused on nonlinearity of the growth process include Fiaschi and Lavezzi (2003), and Aghion, Howitt and Mayer-Foulkes (2005), whereas other research on the emergence of twin peak includes Feyrer (2003).
Ch. 4: From Stagnation to Growth: Unified Growth Theory
257
hypothesis regarding the central roles played by the emergence of human capital formation and the demographic transition in this process. Various qualitative and quantitative unified theories explore plausible mechanisms for the emergence of human capital in the second stage of industrialization and the onset of the demographic transition. These theories focus on the rise in the demand for human capital (due to: technological acceleration, capital-skill complementarity, skilled biased technological change, and reallocation of resources towards skilled-intensive sectors), the decline in child and infant mortality, the rise in life expectancy, the emergence of public education, the decline in child labor, as well as cultural and genetic evolution in the valuation of human capital. The theories suggest that indeed the emergence of human capital formation, and the onset of the demographic transition played a central role in the shift from stagnation to growth. 4.2.1. Alternative mechanisms for the emergence of human capital formation The emergence of human capital formation and its impact on the demographic transition and the technological frontier is a central element in the transition from the Post-Malthusian Regime to the state of sustained economic growth in all unified theories of economic growth in which population, technology and income per capita are endogenously determined.103 Various complementary mechanisms that generate or reinforce the rise in human capital formation have been proposed and examined quantitatively, demonstrating the robustness and the empirical plausibility of this central hypothesis. The rise in the industrial demand for human capital The rise in industrial demand for human capital in advanced stages of industrialization, as documented in Section 2.3.3, and its impact on human capital formation led researchers to incorporate it as a central feature in unified theories of economic growth. The link between industrial development and the demand for human capital have been modeled in various complementary ways. Galor and Weil (2000) modeled the rise in the demand for human capital as an outcome of the acceleration in technological progress, underlying the role of educated individuals in coping with a rapidly changing technological environment. Their mechanism is founded on the premise that the introduction of new technologies increases the demand for skilled labor in the short-run, although in some periods the characteristics of new technologies may be complementarity to unskilled labor, as was the case in the first phase of the Industrial Revolution.104 Subsequent unified theories of economic growth have demonstrated that the rise in the demand for human capital in association with advanced stages of industrialization could emerge from alternative mechanisms, without altering the fundamental insights of 103 Even in the multiple-regime structure of Lucas (2002) a shock to the return to human capital is suggested
in order to generate the switch from the Malthusian Regime to the Modern Growth Regime. 104 Evidence for the complementarity between technological progress (or capital) and skills is provided by Goldin and Katz (1998) and Duffy, Papageorgiou and Perez-Sebastian (2004).
258
O. Galor
the theory. Doepke (2004) constructs his unified theory on the basis of a rising level of skilled-intensive industrial technology, Fernandez-Villaverde (2005) bases his quantitative unified theory on capital-skill complementarity, and Galor and Mountford (2003) generate the rise in the demand for human capital via an increased specialization in the production of skilled-intensive goods, due to international trade. The rise in the demand for human capital stimulated public policy designed to enhance investment in human capital [Galor and Moav (2006)]. In particular, as established in the quantitative unified theory of Doepke (2004), educational policy and child labor laws in England played an important role in human capital formation and the demographic transition. Mortality decline, the rise in life expectancy, and human capital formation Several unified theories of economic growth demonstrate that the basic mechanism for the emergence of human capital proposed by Galor and Weil (2000) can be augmented and reinforced by the incorporation of the effect of the decline in mortality rates and the rise in life expectancy (as documented in Section 2.3.2) on the rise in human capital formation, the decline in the desirable number of surviving offspring, and thus on the transition from stagnation to growth.105 The significant decline in mortality rates in developed countries since the 18th century, as depicted in Figure 24, and the recent decline in mortality rates in less developed countries, as depicted in Figure 25, corresponded to an acceleration in the rise in life expectancy and a significant rise in human capital formation, towards the end of the 19th century in developed countries (Figures 26 and 28), and towards the middle of the 20th century in less developed countries (Figures 27 and 31). The rise in the expected length of the productive life may have increased the potential rate of return to investments in children’s human capital, and thus could have induced an increase in human capital formation along with a decline in fertility. However, despite the gradual rise in life expectancy in developed and less developed countries, investment in human capital has been insignificant as long as the industrial demand for human capital has not emerged. Thus, it appears that the industrial demand for human capital, as documented in Section 2.3.3, provided the inducement for investment in education and the associated reduction in fertility rates, whereas the prolongation of life may have re-enforced and complemented this process. Galor and Weil (1999) argue that the Malthusian interaction between technology and population accelerated the pace of technological progress, improving industrial technology as well as medical and health technologies. Consistent with the historical evidence provided in Section 2.3.3, the improvements in the industrial technology increased the demand for human capital, whereas the development of medical technology and health infrastructure generated a significant rise in life expectancy. The expected rate of return
105 The effect of an increase in life expectancy on the incentive of individuals to invest in their human capital is well established since Ben Porath (1967).
Ch. 4: From Stagnation to Growth: Unified Growth Theory
259
to human capital investment increased therefore due to the prolongation of life, as well as the rise in industrial demand for human capital, enhancing the positive interaction between schooling and technological progress, bringing about a demographic transition and the state of sustained economic growth. Various theories formally examined mechanisms that capture the interaction between human capital formation, the decline in mortality rate, and the rise in life expectancy, in the process of development.106 Cervellati and Sunde (2005) and Boucekkine, de la Croix and Licandro (2003) focus on the plausible role of the reinforcing interaction between life expectancy and human capital formation in the transition from stagnation to growth, abstracting from its effect on fertility decisions. Others suggest that a decline in mortality rates increased the return to investment in human capital via: (a) prolongation of life [Soares (2005)], (b) increased population density and thus the efficiency of the transmission of human capital [Lagerlof (2003a)], (c) increased population growth and the advancement of skill-biased technologies [Weisdorf (2004)], and (d) improved healthiness and thus the capacity to absorb human capital [Hazan and Zoabi (2004)], generating a substitution of quality for quantity, a demographic transition and a transition to a state of sustained economic growth.107 Capital-skill complementarity and the emerging incentives for capitalists to support education reforms The accumulation of physical capital in the early stages of industrialization enhanced the importance of human capital in the production process and generated an incentive for the capitalists to support the provision of public education for the masses.108 Consistent with the evidence provided in Section 2.3.3, Galor and Moav (2006) argue that due to capital-skill complementarity, the accumulation of physical capital by the capitalists in the first phase of the Industrial Revolution increased the importance of human capital in sustaining their rate of return to physical capital, inducing capitalists to support the provision of public education for the masses.109
106 As argued in Section 3.3.1, qualitative and quantitative evidence do not lend credence to the theory that a decline in infant and child mortality rates triggered the decline in the number of surviving offspring and the increase in the investment in offspring’s human capital. 107 See Iyigun (2005) as well. 108 Alternatively, others argued that increased polarization induced the elite to enact costly educational reforms. Grossman and Kim (1999) argue that education decreases predation, and Bowles and Gintis (1975) suggest that educational reforms are designed to sustain the existing social order, by displacing social problems into the school system. In contrast, Bourguignon and Verdier (2000) suggest that if political participation is determined by the education (socioeconomic status) of citizens, the elite may not find it beneficial to subsidize universal public education despite the existence of positive externalities from human capital. 109 Since firms have limited incentive to invest in the general human capital of their workers, in the presence of credit market imperfections, the level of education would be suboptimal unless it would be financed publicly [Galor and Zeira (1993), Durlauf (1996), Fernandez and Rogerson (1996), Benabou (2000), Mookherjee and Ray (2003), and Galor and Moav (2004a)]. Moreover, a mixture of vocational and general education would be enacted [Bertocchi and Spagat (2004)].
260
O. Galor
The decline in child labor Other theories that focused on the transition from stagnation to growth suggest that the central role of human capital formation and the demographic transition can be augmented and reinforced by the incorporation of the adverse effect of the rise in the demand for human capital on child labor. Hazan and Berdugo (2002) suggest that technological change increased the wage differential between parental labor and child labor inducing parents to reduce the number of their children and to further invest in their quality, stimulating human capital formation, a demographic transition, and a shift to a state of sustained economic growth.110 Alternatively, the rise in the importance of human capital in the production process, as documented in Section 2.3.3, induced industrialists to support laws that abolish child labor [Doepke and Zilibotti (2005)], inducing a reduction in child labor, and stimulating human capital formation and a demographic transition. Cultural and genetic evolution of the valuation of human capital Human capital formation and its impact on the decline in the desirable number of surviving offspring may have been reinforced by cultural or genetic evolution in the attitude of individuals towards human capital formation. Consistent with the gradual rise in literacy rates prior to the Industrial Revolution, Galor and Moav (2002) argue that during the epoch of Malthusian stagnation that had characterized most of human existence, individuals with a higher valuation for offspring quality generated an evolutionary advantage and their representation in the population gradually increased. The increase in the rate of return to human capital along with the increase in the bias towards quality in the population reinforced the substitution towards child quality, setting the stage for a significant increase in human capital formation along with a rapid decline in fertility. 4.2.2. Alternative triggers for the demographic transition The demographic transition that separated the Post-Malthusian Regime and the Sustained Growth Regime is a central element in quantitative and qualitative unified theories of economic growth in which population, technology, and income per capita are endogenously determined. As discussed in Section 2.3.2, the demographic transition brought about a reversal in the unprecedented increase in population growth that occurred during the Post-Malthusian Regime, leading to a significant reduction in fertility rates and population growth in various regions of the world, and enabling economies to convert a larger share of the fruits of factor accumulation and technological progress into growth of output per capita.111 The demographic transition enhanced the growth process reducing the dilution of the stock of capital and land, enhancing the investment
110 The decline in the relative wages of children is documented empirically [e.g., Horrell and Humphries
(1995)]. 111 Demographic shocks generate a significant effect on economic growth in Connolly and Peretto (2003) as well.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
261
in the human capital of the population, and altering the age distribution of the population, increasing temporarily the size of the labor force relative to the population as a whole.112 Various complementary mechanisms for the demographic transition have been proposed in the context of unified growth theories, establishing, theoretically and quantitatively the importance of this central hypothesis in the understanding of the transition from stagnation to growth.113 The emergence of human capital formation The gradual rise in the demand for human capital in the process of industrialization, as documented in Section 2.3.3, and its close association with the timing of the demographic transition, has led researchers to argue that the increasing role of human capital in the production process induced households to increase their investment in the human capital of their offspring, eventually leading to the onset of the demographic transition. The link between the rise in the demand for human capital and the demographic transition has been modeled in various complementary ways. Galor and Weil (2000) argue that the gradual rise in the demand for human capital induced parents to invest in the human capital of their offspring. In the early stages of the transition from the Malthusian Regime, the effect of technological progress on parental income permitted the rise in population growth as well as the average population quality. Further increases in the rate of technological progress ultimately induced a reduction in fertility rates, generating a demographic transition in which the rate of population growth declined along with an increase in the average level of education. Thus, consistent with historical evidence, the theory suggests that prior to the demographic transition, population growth increased along with investment in human capital, whereas the demographic transition brought about a decline in population growth along with a further increase in human capital formation. Other theories examine mechanisms that could have reinforced the effect of the rise in the demand for human capital on the demographic transition and the emergence of sustained economic growth, via the decline in benefits from child labor [Hazan and Berdugo (2002), Doepke (2004), and Doepke and Zilibotti (2005)], the decline in mortality rates and the rise in life expectancy [Jones (2001), Lagerlof (2003a), Weisdorf (2004), and Tamura (2004)], and the evolution of preferences for offspring quality [Galor and Moav (2002)], as discussed in Section 3.3. The quantitative examination of Doepke (2004), Fernandez-Villaverde (2005), and Lagerlof (2006) confirm the significance of these channels in originating the demographic transition and the shift from stagnation to growth. 112 Bloom and Williamson (1998) suggest that the cohort effect played a significant role in the growth “miracle” of East Asian countries in the time period 1960–1990. 113 As established in Section 3.3, some mechanisms that were proposed for the demographic transition, such as the decline in infant and child mortality, as well as the rise in income, are inconsistent with the evidence. These mechanisms were therefore excluded in the formulation of unified growth theory.
262
O. Galor
The decline in the gender gap The observed decline in the gender gap in the process of development, as discussed in Section 3.3.4, is an alternative mechanism that could have triggered a demographic transition and human capital formation, as elaborated in other unified theories. A unified theory based upon the decline in the gender wage gap, the associated increase in female labor force participation, and fertility decline was explored by Galor and Weil (1996, 1999), as elaborated in Section 3.3.4. They argue that technological progress and capital accumulation complemented mental intensive tasks and substituted for physical-intensive tasks in the industrial production process. In light of the comparative physiological advantage of men in physical-intensive tasks and women in mental-intensive tasks, the demand for women’s labor input gradually increased in the industrial sector, decreasing monotonically the wage deferential between men and women. In early stages of industrialization, wages of men and women increased, but the rise in female’s relative wages was insufficient to induce a significant increase in women’s labor force participation. Fertility, therefore increased due to the income effect that was generated by the rise in men’s absolute wages. Ultimately, however, the rise in women’s relative wages was sufficient to induce a significant increase in labor force participation, increasing the cost of child rearing proportionally more than households’ income and triggering a demographic transition and a shift from stagnation to growth. Similarly, a transition from stagnation to growth based upon a declining gender gap in human capital formation was proposed by Lagerlof (2003b). He argues that the process of development permitted a gradual improvement in the relative level of female education, raising the opportunity cost of children and initiating a fertility decline and a transition from stagnation to growth.114 4.2.3. Alternative modeling of the transition from agricultural to industrial economy The shift from agriculture to industry that accompanied the transition from stagnation to growth, as described in Section 2.2.3, influenced the specifications of the production structure of most unified theories of economic growth. In some unified theories [e.g., Galor and Weil (2000)] the structure of the aggregate production function and its interaction with technological progress, reflects implicitly a transition from an agricultural to an industrial economy in the process of development. In other theories [e.g., Hansen and Prescott (2002), Kogel and Prskawetz (2001), Hazan and Berdugo (2002),
114 Alternatively, one could have adopted the mechanism proposed by Fernandez, Fogli and Olivetti (2004) and Greenwood, Seshadri and Yorukoglu (2005) for the gradual decline in the education gap and labor force participation between men and women. The former suggests that it reflects a dynamic process in which the home experience of sons of working, educated mothers makes them more likely to prefer educated and working wives, inducing a gradual increase in investment in education as well as in labor force participation among women. The latter suggests that it reflects the reduction in the cost of machines that could substitute for women’s labor at home.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
263
Tamura (2002), Doepke (2004), Galor and Mountford (2003), Bertocchi (2003), and Galor, Moav and Vollrath (2003)] the process of development generates explicitly a transition from an agricultural sector to an industrial sector. In Galor and Weil (2000) production occurs according to a constant returns to scale technology that is subject to endogenous technological progress. The output produced at time t, is Yt = Htα (At X)1−α , where Ht is the aggregate quantity of efficiency units of labor employed in period t, X is land employed in production in every period t, and At represents the endogenously determined technological level in period t. Hence At X are the “effective resources” employed in production in period t. In early stages of development, the economy is agricultural (i.e., the fixed amount of land is a binding constraint on the expansion of the economy). Population growth reduces labor productivity since the rate of technological progress is not sufficiently high to compensate for the land constraint. However, as the rate of technological progress intensifies in the process of development, the economy becomes industrial. Technological progress counterbalances the land constraint, the role of land gradually diminishes, and “effective resources” are expanding at a rate that permit sustained economic growth. Hansen and Prescott (2002) develop a model that captures explicitly the shift from an agricultural sector to an industrial sector in the transition from stagnation to growth. In early stages of development, the industrial technology is not sufficiently productive and production takes place solely in an agricultural sector, where population growth (that is assumed to increase with income) offsets increases in productivity. An exogenous technological progress in the latent constant returns to scale industrial technology ultimately makes the industrial sector economically viable and the economy gradually shifts resources from the agricultural sector to the industrial one. Assuming that the positive effect of income on population is reversed in this transition, the rise in productivity in the industrial sector is not counterbalanced by population growth, permitting the transition to a state of sustained economic growth. Unlike most unified theories in which the time paths of technological progress, population growth, and human capital formation are endogenously determined on the basis on explicit micro-foundations, in Hansen and Prescott (2002) technological progress is exogenous, population growth is assumed to follow the hump-shaped pattern that is observed over human history, and human capital formation (that appears central for the transition) is absent. Based upon this reduced-form approach, they demonstrate that there exists a rate of technological progress in the latent industrial sector, and a well specified reduced-form relationship between population and output, under which the economy will shift from Malthusian stagnation to sustained economic growth. This reduced-form analysis, however, does not advance us in identifying the underlying micro-foundations that led to the transition from stagnation to growth – the ultimate goal of unified growth theory. In accordance with the main hypothesis of Galor and Weil (2000), the transition from stagnation to growth in Hansen and Prescott (2002) is associated with an increase in productivity growth in the economy as a whole. Although productivity growth within each sector is constant, a shift towards the higher productivity growth sector, that is
264
O. Galor
associated with the transition, increases the productivity in the economy, permitting the take-off to a state of sustained economic growth. Moreover, although formally the transition from stagnation to growth in Hansen and Prescott (2002) does not rely on the forces of human capital, if micro-foundations for the critical factors behind the transition would have been properly established, human capital would have played a central role in sustaining the rate of technological progress in the industrial sector and in generating the demographic transition. The lack of an explicit role for human capital in their structure is an artifact of the reduced-form analysis that does not identify the economic factors behind the process of technological change in the latent industrial technology, as well as the forces behind the assumed hump-shaped pattern of population dynamics.115 Thus, Hansen–Prescott’s explicit modeling of the transition from agriculture to industry does not alter the basic insights from the framework of Galor and Weil – a rise in productivity as well as a rise in the demand for human capital is critical for the transition from stagnation to growth. A two-sector framework is instrumental in the exploration of the effect of international trade on the differential timing of the transition from stagnation to growth and the associate phenomenon of the great divergence [Galor and Mountford (2003)], as discussed in Section 6.1. Moreover, a two-sector setting would be necessary in order to examine the incentives of land owners to block education reforms and the process of industrialization [Galor, Moav and Vollrath (2003)], as well as the evolution of property rights and their impact on political reforms [Bertocchi (2003)].
5. Unified evolutionary growth theory 5.1. Human evolution and economic development This section explores the dynamic interaction between human evolution and the process of economic development. It focuses on a recent development of a unified evolutionary growth theory that, based on historical evidence, generates innovative hypotheses about the interplay between the process of development and human evolution, shedding new 115 The demographic pattern assumed by Hansen and Prescott (2002) is critical for the transition from Malthusian stagnation to sustained economic growth. Moreover, human capital appears to be the implicit underlaying force behind their transition as well. In order to generate the features of the Malthusian economy, they set the pace of population growth during this epoch at a level that would generate zero growth rate of output per capita. In the absence of change in this pattern of population growth, output per capita growth is not feasible. Thus, in order to generate output growth along with population growth during the take-off, they assume that at a certain stage the rise in population growth did not fully offset the rise in output (suggesting that parental resources were channeled partly towards child quality in light of a rising demand for human capital). It should be noted that a biological upper bound on the level of fertility that could have generated a take-off mechanically (in an environment characterized by technological change) is inconsistent with the evidence that show that in Western Europe fertility rates continued to increase for nearly a century after the initial take-off.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
265
light on the origin of modern economic growth and the observed intricate evolution of health, life expectancy, human capital, and population growth since the Neolithic Revolution. Unified evolutionary growth theory advances an analytical methodology that is designed to capture the complexity of the dynamic interaction between the economic, social, and behavioral aspects of the process of development and evolutionary processes in the human population. The proposed hybrid between Darwinian methodology and the methodology of unified theories of economic growth permits the exploration of the dynamic reciprocal interaction between the evolution of the distribution of genetic traits and the process of economic development. It captures potential non-monotonic evolutionary processes that were triggered by major socioeconomic transitions, and may have played a significant role in the observed time path of health, life expectancy, human capital, and population growth.116 Humans were subjected to persistent struggle for existence for most of their history. The Malthusian pressure affected the size of the population (as established in Section 2.1.2), and conceivably via natural selection, the composition of the population as well. Lineages of individuals whose traits were complementary to the economic environment generated higher income, and thus a larger number of surviving offspring, and the representation of their traits in the population gradually increased, contributing significantly to the process of development and the take-off from an epoch of Malthusian stagnation to a state of sustained economic growth. Evidence suggests that evolutionary processes in the composition of existing genetic traits may be rather rapid and the time period between the Neolithic Revolution and the Industrial Revolution that lasted about 10,000 years is sufficient for significant evolutionary changes. There are numerous examples of rapid evolutionary changes among various species.117 In particular, evidence establishes that evolutionary changes occurred in the Homo sapiens within the time period that is the focus of the analysis. For instance, lactose tolerance was developed among European and Near Easterners since 116 The conventional methodology of evolutionary stable strategies that has been employed in various fields of economics, ignores the dynamics of the evolutionary process, and is thus inappropriate for the understanding of the “short-run” interaction between human evolution and the process of development since the Neolithic revolution. As will become apparent the dynamics of the evolutionary process are essential for the understanding of the interaction between human evolution and economic growth since the Neolithic revolution – a period marked by fundamental non-monotonic evolutionary processes. 117 The color change that peppered moths underwent during the 19th century is a classic example of evolution in nature [see Kettlewell (1973)]. Before the Industrial Revolution light-colored English peppered moths blended with the lichen-covered bark of trees. By the end of the 19th century a black variant of the moth, first recorded in 1848, became far more prevalent than the lighter varieties in areas in which industrial carbon removed the lichen and changed the background color. Hence, a significant evolutionary change occurred within a time period which corresponds to only hundreds of generations. Moreover, evidence from Daphne Major in the Galapagos suggests that significant evolutionary changes in the distribution of traits among Darwin’s Finches occurred within few generations due to a major drought [Grant and Grant (1989)]. Other evidence, including the dramatic changes in the color patterns of guppies within 15 generations due to changes in the population of predators, are surveyed by Endler (1986).
266
O. Galor
the domestication of dairy animals in the course of the Neolithic revolution, whereas in regions that were exposed to dairy animals in later stages, a larger proportion of the adult population suffers from lactose intolerance. Furthermore, genetic immunity to malaria provided by the sickle cell trait is prevalent among descendants of Africans whose engagement in agriculture improved the breeding ground for mosquitoes and thereby raised the incidence of malaria, whereas this trait is absent among descendants of nearby populations that have not made the transition to agriculture.118 Despite the existence of compelling evidence about the interaction between human evolution and the process of economic development, only few attempts have been made to explore the reciprocal interaction between the process of development and human evolution – an exploration that is likely to revolutionize our understanding of the process of economic development as well as the process of human evolution.119 Galor and Moav (2002) explore the effect of the Malthusian epoch on the evolution of valuation for offspring quality and its role in the transition from stagnation to growth. Ofek (2001) and Saint-Paul (2003) examine the effect of the emergence of markets on the evolution of heterogeneity in the human population. Clark and Hamilton (2003) analyze the relationship between the evolution of time preference and the process of development. Borghans, Borghans and ter-Weel (2004) explore the effect of human cooperation on the evolution of Major Histocompatibility Complex (MHC), and Galor and Moav (2004b) examine the effect of the process of development on the evolution of life expectancy. 5.2. Natural selection and the origin of economic growth The first evolutionary growth theory that captures the interplay between human evolution and the process of economic development in various phases of development, was developed by Galor and Moav (2002). The theory suggests that during the epoch of Malthusian stagnation that had characterized most of human existence, traits of higher valuation for offspring quality generated an evolutionary advantage and their representation in the population gradually increased. This selection process and its effect on investment in human capital stimulated technological progress and initiated a reinforcing interaction between investment in human capital and technological progress that brought about the demographic transition and the state of sustained economic growth.120
118 See Livingston (1958), Weisenfeld (1975) and Durham (1982). 119 The evolution of a wide range of attributes such as time preference, risk aversion, and altruism, in a given
economic environment, has been extensively explored in the economic literature, as surveyed by Bowles (1998) and Robson (2001). 120 The theory is applicable for either social or genetic intergenerational transmission of traits. A cultural transmission is likely to be more rapid and may govern some of the observed differences in fertility rates across regions. The interaction between cultural and genetic evolution is explored by Boyd and Richardson (1985) and Cavalli-Sforza and Feldman (1981), and a cultural transmission of preferences is examined by Bisin and Verdier (2000).
Ch. 4: From Stagnation to Growth: Unified Growth Theory
267
The theory suggests that during the Malthusian epoch, the distribution of valuation for quality lagged behind the evolutionary optimal level. The evolution of the human brain in the transition to Homo sapiens and the complementarity between brain capacity and the reward for human capital has increased the evolutionary optimal investment in the quality of offspring (i.e., the level that maximizes reproduction success).121 Moreover, the increase in the return to human capital in the aftermath of the Neolithic Revolution increased the evolutionary optimal level of investment in child quality. The agricultural revolution facilitated the division of labor and fostered trade relationships across individuals and communities, enhancing the complexity of human interaction and raising the return to human capital. Thus, individuals with traits of higher valuation for offspring’s quality generated higher income and, in the Malthusian epoch when child rearing was positively affected by aggregate resources, a larger number of offspring. Traits of higher valuation for quality gained the evolutionary advantage and their representation in the population increased over time. The Malthusian pressure increased the representation of individuals whose preferences are biased towards child quality, positively affecting investment in human capital and ultimately the rate of technological progress. In early stages of development, the proportion of individuals with higher valuation for quality was relatively low, investment in human capital was minimal, resources above subsistence were devoted primarily to child rearing, and the rate of technological progress was rather slow. Technological progress therefore generated proportional increases in output and population and the economy was in the vicinity of a Malthusian equilibrium, where income per capita is constant, but the proportion of individuals with high valuation for quality was growing over time.122 As the fraction of individuals with high valuation for quality continued to increase, technological progress intensified, raising the rate of return to human capital. The increase in the rate of technological progress generated two effects on the size and the quality of the population. On the one hand, improved technology eased households’ budget constraints and provided more resources for quality as well as quantity of children. On the other hand, it induced a reallocation of these increased resources toward child quality. In the early stages of the transition from the Malthusian Regime, the
121 The evolutionary process in valuation for quality that was triggered by the evolution of the human brain has not reached a new evolutionary stable state prior to the Neolithic period because of the equality that characterized resource allocation among hunter-gatherers tribes. Given this tribal structure, a latent attribute of preferences for quality, unlike observable attributes such as strength and intelligence, could not generate a disproportionate access to sexual mates and resources that could affect fertility rates and investment in offspring’s quality, delaying the manifestation of the potential evolutionary advantage of these traits. It was the emergence of the nuclear family in the aftermath of the agricultural revolution that fostered intergenerational links, and thereby enhanced the manifestation of the potential evolutionary advantage of this trait. 122 Unlike Galor and Weil (2000) in which the adverse effect of limited resources on population growth delays the process of development, in the proposed theory the Malthusian constraint generates the necessary evolutionary pressure for the ultimate take-off.
268
O. Galor
effect of technological progress on parental income dominated, and the rate of population growth as well as the average quality increased, further accelerating technological progress. Ultimately, the rate of technological progress induced a universal investment in human capital along with a reduction in fertility rates, generating a demographic transition in which the rate of population growth declined along with an increase in the average level of education. The positive feedback between technological progress and the level of education reinforced the growth process, setting the stage for the transition to a state of sustained economic growth.123 During the transition from the Malthusian epoch to the sustained growth regime, once the economic environment improved sufficiently, the significance of quality for survival declined, and traits of higher valuation for quantity gained the evolutionary advantage. Namely, as technological progress brought about an increase in income, the Malthusian pressure relaxed and the domination of wealth in fertility decisions diminished. The inherent advantage of higher valuation for quantity in reproduction has started to dominate, and individuals whose preferences are biased towards child quantity gained the evolutionary advantage. Nevertheless, the growth rate of output per worker has remained positive since the high rate of technological progress sustained an attractive return to investment in human capital even from the viewpoint of individuals whose valuation for quality is relatively low. The transition from stagnation to growth is an inevitable by-product of the interaction between the composition of the population and the rate of technological progress in the Malthusian epoch. However, for a given composition of population, the timing of the transition may differ significantly across countries and regions due to historical accidents, as well as variation in geographical, cultural, social and institutional factors, trade patterns, colonial status, and public policy. 5.2.1. Primary ingredients The theory is based upon the interaction between several building blocks: the Darwinian elements, the Malthusian elements, the nature of technological progress, the determinants of human capital formation, and the factors that affect parental choice regarding the quantity and quality of offspring. The Darwinian elements. The theory incorporates the main ingredients of Darwinian evolution (i.e., variety, intergenerational transmission of traits, and natural selection) into the economic environment. Inspired by fundamental components of the Darwinian theory [Darwin (1859, 1871)], individuals do not operate consciously so as to assure their evolutionary advantage. Nevertheless, their preferences (or strategies) assure that 123 The theory suggests that waves of rapid technological progress in the Pre-Industrial Revolution era did not generate sustained economic growth due to the shortage of preferences for quality in the population. Although in these previous episodes technological progress temporarily increased the return to quality, the level of human capital that was generated by the response of the existing population to the incentive to invest in human capital, was insufficient to sustain technological progress and economic growth.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
269
those individuals whose operations are most complementary to the environment would eventually dominate the population. Individuals’ preferences are defined over consumption above subsistence as well as over the quality and the quantity of their children.124 These preferences capture the Darwinian survival strategy as well as the most fundamental trade-offs that exist in nature – the trade-off between the resources allocated to the parent and the offspring, and the trade-off between the number of offspring and resources allocated to each offspring.125 The economy consists of a variety of types of individuals distinguished by the weight given to child quality in their preferences.126 This trait is assumed to be transmitted intergenerationally. The economic environment determines the type with the evolutionary advantage (i.e., the type characterized by higher fertility rates), and the distribution of preferences in the population evolves over time due to differences in fertility rates across types.127 The significance that individuals attribute to child quantity as well as to child quality reflects the well-known variety in the quality-quantity survival strategies (or in the K and r strategies) that exists in nature [e.g., MacArthur and Wilson (1967)]. Human beings, like other species, confront the basic trade-off between offspring’s quality and quantity in their implicit Darwinian survival strategies. Although a quantity-biased preference has a positive effect on fertility rates and may therefore generate a direct evolutionary advantage, it adversely affects the quality of offspring, their income, and their fitness and may therefore generate an evolutionary disadvantage. “Increased bearing is bound to be paid for by less efficient caring” [Dawkins (1989, p. 116)]. As was established in the evolutionary biology literature since the seminal work of Lack (1954), the allocation of resources between offspring “caring” and “bearing” is subjected to evolutionary changes.128 124 The subsistence consumption constraint is designed to capture the fact that the physiological survival of the parent is a pre-condition for the survival of the lineage (dynasty). Resources allocated to parental consumption beyond the subsistence level may be viewed as a force that rises parental productivity and resistance to adverse shocks (e.g., famine and disease), generating a positive effect on the fitness of the parent and the survival of the lineage. This positive effect, however, is counterbalanced by the implied reduction in resources allocated to the offspring, generating a negative effect on the survival of the lineage. 125 Resources allocated to quality of offspring in different stages of development take different forms. In early stages of development it is manifested in investment in the durability of the offspring via better nourishment and parental guidance, whereas in mature stages, investment in quality may capture formal education. 126 The analysis abstracts from heterogeneity in the degree of the trade-off between resources allocated to parent and offspring. The introduction of this element would not alter the qualitative results. 127 Recent research across historical and modern data from the United States and Europe suggests that fertility behavior has a significant hereditary component [Rodgers et al. (2001a)]. For instance, as established recently by Kohler et al. (1999) and Rodgers et al. (2001b), based on the comparison of fertility rates among identical and fraternal twins born in Denmark during the periods 1870–1910 and 1953–1964, slightly more than one-quarter of the variance in completed fertility is attributable to genetic influence. These findings are consistent with those of Rodgers and Doughty (2000) based on kinship data from the United States. 128 Lack (1954) suggests that clutch sizes (i.e., number of eggs per nest), among owls and other predatory vole-eating birds, for instance, are positively related to food abundance. He argues that the clutch size is
270
O. Galor
The Malthusian elements. Individuals are subjected to subsistence consumption constraint and as long as the constraint is binding, an increase in income results in an increase in population growth along with an increase in the average quality of a minor segment of the population. Technological progress, which brings about temporary gains in income per capita, triggers therefore an increase in the size of the population that offsets the gain in income per capita due to the existence of diminishing returns to labor. Growth in income per capita is generated ultimately, despite decreasing returns to labor, since technological progress induces investment in human capital among a growing minority. The determinants of technological progress. The average level of human capital as reflected by the composition of the population is the prime engine of technological progress.129 The origin of human capital formation. Technological change raises the demand for human capital. Technological progress reduces the adaptability of existing human capital for the new technological environment and educated individuals (and thus offspring of parent with high valuation for quality) have a comparative advantage in adapting to the new technological environment.130 The determination of paternal decision regarding offspring quantity and quality. Individuals choose the number of children and their quality based upon their preferences for quality as well as their time constraint.131 The rise in the (genetic or cultural) bias
selected such that under any feeding conditions fertility rates ensure the maximal reproductive success. Furthermore, Cody (1966) documents the existence of significant differences between clutch sizes of the same bird species on islands and nearby mainland localities of the same latitude. In temperate regions where food is more abundant in the mainland than on islands, the average clutch size is smaller on the islands. For instance, for Cyanoramphus novaezelandeae, the average mainland clutch is 6.5 whereas the average in the island is 4. 129 This link between education and technological change was proposed by Nelson and Phelps (1966) and is supported empirically by Easterlin (1981), Doms et al. (1997), as well as others. In order to focus on the role of the evolutionary process, the model abstracts from the potential positive effect of the size of the population on the rate of technological progress. Adding this scale effect would simply accelerate the transition process [e.g., Galor and Weil (2000)]. Consistently with Mokyr (2002) who argues that the effect of human capital accumulation on technological progress becomes significant only in the course of the Scientific Revolution that preceded the Industrial Revolution, the effect of human capital accumulation on the rate of technological progress need not be significant prior to the scientific revolution, as long as it becomes significant prior to the Industrial Revolution. 130 See Schultz (1964) and Nelson and Phelps (1966). If the return to education rises with the level of technology rather than with the rate of technological progress, the qualitative analysis would not be affected. However, this alternative would imply that changes in technology were skill-biased throughout human history in contrast to those periods in which technological change was skilled-saving, notably, in the first phase of the Industrial Revolution. 131 Anthropological evidence suggests that fertility control was indeed exercised even prior to the Neolithic Revolution. Reproductive control in hunter-gatherer societies is exemplified by “pacing birth” (e.g., birth every four years) conducted by tribes who live in small, semi nomadic bands in Africa, Southeast Asia, and New Guinea in order to prevent the burden of carrying several children while wandering. Similarly, Nomadic women of the Kung use no contraceptives but nurse their babies frequently, suppressing ovulation and menstruation for two to three years after birth, and reaching a mean interval between births of 44 months.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
271
towards quality, as well as the rise in the demand for human capital, induce parents to substitute quality for quantity of children. 5.2.2. Main hypotheses and their empirical assessment The theory generates several hypotheses about human evolution and the process of development, underlying the role of natural selection in: (i) the gradual process of human capital formation and thus technological progress prior to the Industrial Revolution, and (ii) the acceleration of the interaction between human capital and technological progress in the second phase of the Industrial Revolution, the associated demographic transition, and the emergence of a state of sustained economic growth. The main hypotheses (H1) During the initial phases of the Malthusian epoch, the growth rate of output per capita is nearly zero and the growth rate of population and literacy rates is minuscule, reflecting the sluggish pace of technological progress, the low representation of individuals with high valuation for child quality, and the slow pace of the evolutionary process. This hypothesis is consistent with the characteristics of the Malthusian epoch, as described in Section 2.1. (H2) In the pre-demographic transition era, traits for higher valuation for offspring quality generated an evolutionary advantage. Namely, individuals with higher valuation for the quality of children had a larger number of surviving offspring and their representation in the population increased over time. In contrast, in the post-demographic transition era, when income per capita has no longer been the binding constraint on fertility decisions, individuals with higher valuation for offspring quantity have an evolutionary advantage, bearing a lager number of surviving offspring. Thus, in the pre-demographic transition era, the number of surviving offspring was affected positively by parental education and parental income whereas in the post-demographic transition era this pattern is reversed and more educated, higher income individuals have a smaller number of surviving offspring. Clark and Hamilton (2003) examine empirically this hypothesis on the basis of data from wills written in England in the time period 1620–1636. The wills were written in a closed proximity to the death of a person, in urban and rural areas, and across a large variety of occupations and wealth. They contain information about the number of surviving offspring, literacy of testator (measured by whether the will was signed), occupation of testator (if male), the amount of money bequeathed and to whom (spouse, children, the poor, unrelated persons), and houses and land that were bequeathed. Based on this data, Clark and Hamilton find a positive and statistically significant effect of liter-
272
O. Galor
acy (and wealth) on the number of surviving offspring.132 They confirm the hypothesis that literate people (born, according to the theory, to parents with quality-bias) had an evolutionary advantage in this (pre-demographic transition) period.133 The negative relationship between education and fertility within a country in the post-demographic transition era was documented extensively.134 (H3) The increased the representation of individuals with higher valuation for quality, gradually increased the average level of investment in human capital,135 permitting a slow growth of output per capita. The prediction about the rise in human capital prior to the Industrial Revolution is consistent with historical evidence. Various measures of literacy rates demonstrate a significant rise in literacy rates during the two centuries that preceded the Industrial Revolution in England.136 As depicted in Figure 42, male literacy rates increased gradually in the time period 1600–1760. Literacy rates for men doubled over this period, rising from about 30% in 1600 to over 60% in 1760. Similarly, as reported by Cipolla (1969), literacy rates of women more than tripled from less than 10% in 1640 to over 30% in 1760.137 Moreover, as argued by Clark (2003), human capital accumulation in England began in an era when the market rewards for skill acquisition were at historically low levels, consistent with the argument that the rise in human capital reflected a rise in the preference for quality offspring.
132 In addition, Boyer (1989) argues that in early 19th century England, agricultural laborers’ income had a positive effect on fertility: birth rates increased by 4.4% in response to a 10% increase in annual income. Further evidence is surveyed by Lee (1997). 133 Interestingly, in New France, where land was abundant, and thus fertility decisions were not constraint by the availability of resources, the number of surviving offspring was higher among less educated individuals. These findings are consistent with the theory as well. If resource constraint is not binding for fertility decisions (e.g., in the post-demographic transition era, or due to a positive shock to income in the Malthusian era), individuals with higher valuation for quantity gain an evolutionary advantage. 134 See, for instance, Kremer and Chen (2002). 135 In contrast to Galor and Weil (2000) in which the inherent positive interaction between population and technology during the Malthusian Regime is the force behind the increase in the rate of technological progress that induced investments in human capital and led to further technological progress, a demographic transition, and sustained economic growth, Galor and Moav (2002) is structured such that the gradual change in the composition of the population (rather than the size of the population) brings about the take-off from stagnation to growth. Thus, a scale effect is not needed for the take-off. However, this is just a simplifying modeling devise and both forces could operate simultaneously in triggering the take-off. 136 Moreover, this hypothesis appears consistent with the increase in the number and size of universities in Europe since the establishment of the first university in Bologna in the 11th century, significantly outpacing the growth rate of population. 137 This pattern is robust and is observed in various dioceses over this period. For instance Cressy (1981, Table 6.3, p. 113) reports a gradual rise in average literacy rate of average of yeomen, husbandmen and tradesmen in Norwich from 30% in 1580 to nearly 61% in 1690, and Cressy (1980, Table 7.1, p. 143) reports a gradual rise in Gentle literacy in the diocese of Durham between 1565 and 1624.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
273
Figure 42. The rise in male literacy rates prior and during the industrial revolution: England: 1600–1900. Sources: Cipolla (1969), Stone (1969), and Schofield (1973).
(H4) The acceleration in the rate of technological progress that was reinforced by the investment in human capital of individuals with high valuation for offspring quality, increased the demand for human capital in the later part of the Post-Malthusian Regime, generating a universal investment in human capital, a demographic transition and a rapid pace of economic growth. The hypothesis is consistent with the evidence, provided in Section 2.3 and depicted partly in Figure 41, about the significant rise in the industrial demand for human capital in the second phase of the Industrial Revolution, the marked increased in educational attainment, the emergence of universal education towards the end of the 19th century in association with a decline in fertility rates, and a transition to a state of sustained economic growth. 5.3. Complementary mechanisms The theory suggests that during the Malthusian epoch hereditary human traits, physical or mental, that generate higher earning capacity, and thereby potentially larger number of offspring, would generate an evolutionary advantage and would dominate the population. Hereditary traits that stimulate technological progress or raise the incentive to invest in offspring’s human capital (e.g., ability, longevity, and a preference for quality), may trigger a positive feedback loop between investment in human capital and technological progress that would bring about a take-off from an epoch of Malthusian stagnation, a demographic transition, and a shift to a state of sustained economic growth. Hence, the struggle for existence that had characterized most of human history stimulated natural selection and generated an evolutionary advantage to individuals whose
274
O. Galor
characteristics are complementary to the growth process, eventually triggering a takeoff from an epoch of stagnation to sustained economic growth. Galor and Moav (2002) focus on the evolution of the trade-off between resources allocated to the quantity and the quality of offspring. Their framework of analysis can be modified to account for the interaction between economic growth and the evolution of other hereditary traits. 5.3.1. The evolution of ability and economic growth Suppose that individual’s preferences are defined over consumption above a subsistent level and over child quality and quantity. Individuals are identical in their preferences, but differ in their hereditary innate ability. Suppose further that offspring’s level of human capital is an increasing function of two complementary factors: innate ability and investment in quality. Thus, since the marginal return to investment in child quality increases with ability, higher-ability individuals and hence dynasties would allocate a higher fraction of their resources to child quality. In the Malthusian era, individuals with a higher ability generate more income and hence are able to allocate more resources for child quality and quantity. High ability individuals, therefore, generate higher income due to fact that their innate ability as well as their quality are higher. In the Malthusian era fertility rates are positively affected by the level of income and (under plausible configurations) the high ability individuals have therefore an evolutionary advantage over individuals of lower ability. As the fraction of individuals of the high ability type increases, investment in quality rises, and technological progress intensifies. Ultimately the dynamical system changes qualitatively, the Malthusian temporary steady state vanishes endogenously and the economy takes off from the Malthusian trap. Once the evolutionary process generates a positive feedback between the rate of technological progress and the level of education, technological progress is reinforced, the return to human capital increases further, setting the stage for the demographic transition and the shift to a state of sustained economic growth. 5.3.2. The evolution of life expectancy and economic growth Suppose that individuals differ in their level of health due to hereditary factors. Suppose further that there exists a positive interaction between the level of health and economic well-being. Higher income generates a higher level of health, whereas higher level of health increases labor productivity and life expectancy. Parents that are characterized by high life expectancy, and thereby expect their offspring to have a longer productive life, would allocate more resources toward child quality. In the Malthusian era, fertility rates are positively affected by the level of income and individuals with higher life expectancy, and therefore higher quality and higher income, would have (under plausible configurations) an evolutionary advantage. Natural selection therefore, increases the level of health as well as the quality of the population. Eventually, this process generates a positive feedback loop between investment in child quality, technological
Ch. 4: From Stagnation to Growth: Unified Growth Theory
275
progress and health, bringing about a transition to a sustained economic growth with low fertility rates and high longevity. Alternatively, Galor and Moav (2004b) develop an evolutionary growth theory that captures the intricate time path of life expectancy in the process of development, shedding new light on the origin of the remarkable rise in life expectancy since the Agricultural Revolution. The theory argues that social, economic and environmental changes that were associated with the transition from hunter-gatherer tribes to sedentary agricultural communities, and ultimately to urban societies, affected the nature of the environmental hazards confronted by the human population, triggering an evolutionary process that had a significant impact on the time path of human longevity. The theory suggests that the deterioration in the health environment enhanced the genetic potential for longer life expectancy thereby playing a significant role in the dramatic impact of recent improvements in health infrastructure on the prolongation of human life. The rise in population density, the domestication of animals, and the increase in work effort in the course of the Agricultural Revolution, increased the exposure and the vulnerability of humans to environmental hazards such as infectious diseases and parasites, increasing the extrinsic mortality risk and leading to the observed temporary decline in life expectancy during the Neolithic period.138 The theory suggests, however, that the evolutionary optimal allocation of parental resources towards somatic investment, repairs, and maintenance (e.g., enhanced immune system, DNA repairs, accurate gene regulation, tumor suppression, and antioxidants) was altered in the face of the fundamental trade-off between current and future reproduction. The rise in the extrinsic morality risk generated an evolutionary advantage to individuals who were genetically pre-disposed towards higher somatic investment leading to the observed increase in life expectancy in the post-Neolithic period. Galor and Moav (2004b) suggest, therefore, that the increase in the extrinsic morality risk (i.e., risk associated with environmental factors) in the course of the Agricultural Revolution triggered an evolutionary process that gradually altered the distribution of genes in the human population that are associated with the intrinsic mortality risk (i.e., physiological and biochemical decay over lifetime). Individuals that were characterized by a higher genetic predisposition towards somatic investment, repairs, and maintenance gained the evolutionary advantage during this transition, and their representation in the population increased over time.139 Despite the increase in the extrinsic mortality risk 138 Most comparisons between hunter-gatherers and farmers [e.g., Cohen (1989)] suggest that, in the same locale, farmers suffered higher rates of infection due to the increase in human settlements, poorer nutrition due to reduced meat intake and greater interference with mineral absorption by the cereal-based diet. Consequently, Neolithic farmers were shorter and had a lower life expectancy relative to Mesolithic hunter-gatherers. Although it is difficult to draw reliable conclusion about relative life expectancy in these periods, because skeletal samples are often distorted and incomplete, available evidence suggests that prehistoric hunter-gatherers often fared relatively well in comparison to later populations, particularly with reference to the survival of children. The Illinois Valley provides life tables for hunter-gatherers which confirm the fact that their life expectancies matched or exceeded those of later groups. 139 For the effect of somatic maintenance system on longevity see Kirkwood (1998).
276
O. Galor
that brought about a temporary decline in life expectancy, longevity eventually increased beyond the peak that existed in the hunter-gatherer society, due to the changes in the distribution of genes in the human population. This evolutionary process in life expectancy reinforced the interaction between investment in human capital, life expectancy, and technological progress thereby expediting the demographic transition and enhancing the economic transition from stagnation to growth.140 Moreover, the biological upper bound of longevity gradually increased, generating the biological infrastructure that contributed significantly to the impact of recent improvements in medical technology on the dramatic prolongation of life expectancy.
6. Differential takeoffs and the great divergence The last two centuries have witnessed dramatic changes in the distribution of income and population across the globe. The differential timing of the take-off from stagnation to growth across countries and the corresponding variations in the timing of the demographic transition have led to a great divergence in income, as depicted in Figure 32, and to significant changes in the distribution of population around the globe, as depicted in Figure 33. Some regions have excelled in the growth of income per capita, while other regions have been dominant in population growth. Inequality in the world economy had been insignificant until the 19th century. The ratio of GDP per capita between the richest region and the poorest region in the world was only 1.1 : 1 in 1000 AD, 2 : 1 in 1500 and 3 : 1 in 1820. In contrast, the past two centuries have been characterized by a ‘Great Divergence’ in income per capita among countries and regions. In particular, the ratio of GDP per capita between the richest and the poorest regions has widened considerably from a modest 3 : 1 ratio in 1820, to a large 18 : 1 ratio in 2001. An equally impressive transformation occurred in the distribution of world population across regions, as depicted in Figure 33. The earlier take-off of Western European countries generated a 16% increase in the share of their population in the world economy within the time period 1820–1870. However, the early onset of the Western European demographic transition, and the long delay in the demographic transitions of less developed regions, well into the second half of the 20th century, led to a 55% decline in the share of Western European population in the world in the time period 1870–1998. In contrast, the prolongation of the Post-Malthusian period of less developed regions and the delay in their demographic transitions, generated a 84% increase in Africa’s share of world population, from 7% in 1913 to 12.9% in 1998, an 11% increase in Asia’s share of world population from 51.7% in 1913 to 57.4% in 1998, and a four-fold increase in Latin American’s share in world population from 2% in 1820 to 8.6% in 1998.
140 The evolution of the human brain along with the evolution of life expectancy prior to the Neolithic revolution is examined by Robson and Kaplan (2003).
Ch. 4: From Stagnation to Growth: Unified Growth Theory
277
The phenomenon of the Great Divergence in income per capita across regions of the world over the past two centuries, that was associated with the take-off from the epoch of near stagnation to a state of sustained economic growth, presents intriguing questions about the growth process. How does one account for the sudden take-off from stagnation to growth in some countries in the world and the persistent stagnation in others? Why has the positive link between income per capita and population growth reversed its course in some economies but not in others? Why have the differences in per capita incomes across countries increased so markedly over the last two centuries? Has the transition to a state of sustained economic growth in advanced economies adversely affected the process of development in less-developed economies? 6.1. Non-unified theories The origin of the Great Divergence has been a source of controversy. The relative roles of geographical and institutional factors, human capital formation, ethnic, linguistic, and religious fractionalization, colonialism and globalization have been at the center of a debate about the origins of this remarkable change in the world income distribution in the past two centuries. The role of institutional and cultural factors has been the focus of influential hypotheses regarding the origin of the great divergence. North (1981), Landes (1998), Mokyr (1990, 2002), Hall and Jones (1999), Parente and Prescott (2000), and Acemoglu, Johnson and Robinson (2002) have argued that institutions that facilitated the protection of property rights and enhanced technological research and the diffusion of knowledge, have been the prime factors that enabled the earlier European take-off and the great technological divergence across the globe.141 The effect of geographical factors on economic growth and the great divergence have been emphasized by Jones (1981), Diamond (1997) and Gallup, Sachs and Mellinger (1998).142 The geographical hypothesis suggests that advantageous geographical conditions made Europe less vulnerable to the risk associated with climate and diseases, leading to the early European take-off, whereas adverse geographical conditions in disadvantageous regions (e.g., harsh climate, prevalence of diseases, scarcity of natural resources, high transportation costs, limited regional diffusion of knowledge and technology), generated permanent hurdles for the process of development, contributing to the great divergence.143 The exogenous nature of the geographical factors and the endogenous nature of the institutional factors lead researchers to hypothesize that initial geographical conditions 141 Barriers to technological adoption that may lead to divergence are explored by Caselli and Coleman
(2002), Howitt and Mayer-Foulkes (2005) and Acemoglu, Aghion and Zilibotti (2004). 142 See also Hall and Jones (1999), Masters and McMillan (2001) and Hibbs and Olson (2005). 143 Bloom, Canning and Sevilla (2003) cross section analysis rejects the geographical determinism, but maintains nevertheless that favorable geographical conditions have mattered for economic growth since they increase the likelihood of an economy to escape a poverty trap. See Przeworski (2003) as well.
278
O. Galor
had a persistent effect on the quality of institutions, leading to divergence and overtaking in economic performance. Engerman and Sokoloff (2000) provide descriptive evidence that geographical conditions that led to income inequality, brought about oppressive institutions designed to preserve the existing inequality, whereas geographical characteristics that generated an equal distribution of income led to the emergence of growth promoting institutions. Acemoglu, Johnson and Robinson (2002) provide evidence that reversals in economic performance across countries have a colonial origin, reflecting institutional reversals that were introduced by European colonialism across the globe.144 “Reversals of fortune” reflect the imposition of extractive institutions by the European colonialists in regions where favorable geographical conditions led to prosperity, and the implementation of growth enhancing institutions in poorer regions.145 Furthermore, the role of ethnic, linguistic, and religious fractionalization in the emergence of divergence and “growth tragedies” has been linked to their effect on the quality of institutions. Easterly and Levine (1997) and Alesina et al. (2003) demonstrate that geopolitical factors brought about a high degree of fractionalization in some regions of the world, leading to the implementation of institutions that are not conducive for economic growth and thereby to diverging growth paths across regions. The role of human capital in the great divergence is underlined in the unified growth theories of Galor and Weil (2000), Galor and Moav (2002), Doepke (2004), FernandezVillaverde (2005), Lagerlof (2006), as well as others. These theories establish theoretically and quantitatively that the rise in the technologically-driven demand for human capital in the second phase of industrialization and its effect on human capital formation and on the onset of the demographic transition have been the prime forces in the transition from stagnation to growth and thus in the emergence of the associated phenomena of the great divergence. In particular, they suggests that once the technologically-driven demand for human capital emerged in the second phase of industrialization, the prevalence of human capital promoting institutions determined the extensiveness of human capital formation, and therefore the rapidity of technological progress, the timing of the demographic transition, the pace of the transition from stagnation to growth, and thus the distribution of income in the world economy. Empirical research is inconclusive about the significance of human capital rather than institutional factors in the process of development. Some researchers suggest that initial geographical conditions affected the current economic performance primarily via their effect on institutions. Acemoglu, Johnson and Robinson (2002), Easterly and Levine (2003), and Rodrik, Subramanian and Trebbi (2004) provide evidence that variations in the contemporary growth processes across countries can be attributed to institutional factors whereas geographical factors are secondary, operating primarily via variations in institutions. 144 Additional aspects of the role of colonialism in comparative developments are analyzed by Bertocchi and
Canova (2002). 145 Brezis, Krugman and Tsiddon (1993) attribute technological leapfrogging to the acquired comparative advantage (via learning by doing) of the current technological leaders in the use of the existing technologies.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
279
Glaeser et al. (2004) revisit the debate whether political institutions cause economic growth, or whether, alternatively, growth and human capital accumulation lead to institutional improvement. In contrast to earlier studies, they find that human capital is a more fundamental source of growth than are the institutions. Moreover, they argue that poor countries emerge from poverty through good policies (e.g., human capital promoting policies) and only subsequently improve their political institutions. A theory that unifies the geographical and the human capital paradigms, capturing the transition from the domination of the geographical factors in the determination of productivity in early stages of development, to the domination of human capital promoting institutions in mature stages of development has been proposed by Galor, Moav and Vollrath (2003). The theory identifies and establishes the empirical validity of a channel through which favorable geographical conditions, that were inherently associated with inequality, affected the emergence of human capital promoting institutions (e.g., public schooling, child labor regulations, abolishment of slavery, etc.), and thus the pace of the transition from an agricultural to an industrial society.146 They suggest that the distribution of land within and across countries affected the nature of the transition from an agrarian to an industrial economy, generating diverging growth patterns across countries. The accumulation of physical capital in the process of industrialization raised the importance of human capital in the growth process, reflecting the complementarity between capital and skills. Investment in human capital, however, was sub-optimal due to credit market imperfections, and public investment in education was growth-enhancing. Nevertheless, human capital accumulation did not benefit all sectors of the economy. Due to a low degree of complementarity between human capital and land, universal public education increased the cost of labor beyond the increase in average labor productivity in the agricultural sector, reducing the return to land. Landowners, therefore, had no economic incentives to support these growth enhancing educational policies as long as their stake in the productivity of the industrial sector was insufficient. Land abundance, which was beneficial in early stages of development, brought about a hurdle for human capital accumulation and economic growth among countries that were marked by an unequal distribution of land ownership.147 6.2. Unified theories Unified theories of economic growth generate direct hypotheses about the factors that determine the timing of the transition from stagnation to growth and thus the causes of the Great Divergence. The timing of the transition may differ significantly across 146 As established by Chanda and Dalgaard (2003), variations in the structural composition of economies
and in particular the allocation of scarce inputs between the agricultural and the non-agricultural sectors are important determinants of international differences in TFP, accounting for between 30% and 50% of these variations. 147 Berdugo, Sadik and Sussman (2003) explore an alternative theory of divergence and overtaking that links natural resources abundance, the quality of learning institutions, and retardation in technology adoption.
280
O. Galor
countries and regions due to historical accidents, as well as variations in geographical, cultural, political, social and institutional factors that affected the vital interaction between population and technology in the Malthusian epoch, and the fundamental links between technological progress, human capital formation, and the demographic transition, in the Post-Malthusian Regime as well as in the Modern Growth Regime.148 6.2.1. Human capital promoting institutions The role of human capital in the take-off from stagnation to growth and thus in the great divergence was underlined in the unified theories of Galor and Weil (2000), Galor and Moav (2002), Doepke (2004), Fernandez-Villaverde (2005), Lagerlof (2006), as well as others, as explored in Section 4. These theories establish theoretically and quantitatively that the rise in the demand for human capital in the second phase of industrialization, and its effect on human capital formation, and the onset of the demographic transition that swept the world in the course of the last century, have been the prime forces in the transition from stagnation to growth and thus in the emergence of the associated phenomena of the Great Divergence. Furthermore, they suggest that variations in the timing of the transition from stagnation to growth (e.g., England’s earlier industrialization in comparison to China), and thus differences in economic performance across countries, may reflect initial differences in geographical factors and historical accidents and their manifestation in variations in institutional, demographic, and cultural factors, trade patterns, colonial status, and public policy. Consistently with the findings of Glaeser et al. (2004), these unified theories suggest that once the technologically-driven demand for human capital emerged in the second phase of industrialization, the prevalence of human capital promoting institutions determined the swiftness of human capital formation, the timing of the demographic transition, the pace of the transition from stagnation to growth, and thereby the observed distribution of income in the world economy. 6.2.2. Globalization and the great divergence This subsection explores a unified growth theory that generates a transition from stagnation to growth along with a great divergence, focusing on the asymmetric effect of globalization on the timing of the transition of developed and less developed countries from Malthusian stagnation to sustained economic growth. Galor and Mountford
148 Related to the unified paradigm, Pomeranz (2000) suggest that the discovery of the New World, enabled Europe, via Atlantic trade, to overcome ‘land constraints’ and to take off technologically. The inflow of grain and other commodities as well as the outflow of migrants during the 19th century may have played a crucial role in Europe’s development. By easing the land constraint at a critical point – when income per capita had begun to rise rapidly, but before the demographic transition had gotten under way – the “ghost acres” of the New World provided a window of time which allowed Europe to pull decisively away from the Malthusian equilibrium.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
281
(2003) suggest that sustained differences in income and population growth across countries could be attributed to the contrasting effect of international trade on industrial and non-industrial nations. Consistent with the evidence provided in Section 2, their theory suggests that the expansion of international trade in the 19th century and its effect on the pace of individualization has played a major role in the timing of demographic transitions across countries and has thereby been a significant determinant of the distribution of world population and a prime cause of the ‘Great Divergence’ in income levels across countries in the last two centuries. International trade had an asymmetrical effect on the evolution of industrial and non-industrial economies. While in the industrial nations the gains from trade were directed primarily towards investment in education and growth in output per capita, a significant portion of the gains from trade in non-industrial nations was channeled towards population growth.149 In the second phase of the Industrial Revolution, international trade enhanced the specialization of industrial economies in the production of industrial, skilled intensive, goods. The associated rise in the demand for skilled labor induced a gradual investment in the quality of the population, expediting a demographic transition, stimulating technological progress and further enhancing the comparative advantage of these industrial economies in the production of skilled intensive goods. In non-industrial economies, in contrast, international trade has generated an incentive to specialize in the production of unskilled intensive, non-industrial, goods. The absence of significant demand for human capital has provided limited incentives to invest in the quality of the population, and the gains from trade have been utilized primarily for a further increase in the size of the population, rather than in the income of the existing population. The demographic transition in these non-industrial economies has been significantly delayed, increasing further their relative abundance of unskilled labor, enhancing their comparative disadvantage in the production of skilled intensive goods, and delaying their process of development. The research suggests, therefore, that international trade affected persistently the distribution of population, skills, and technologies in the world economy, and has been a significant force behind the ‘Great Divergence’ in income per capita across countries.150 149 In contrast to the recent literature on the dynamics of comparative advantage [e.g., Findlay and Keirzkowsky (1983), Grossman and Helpman (1991), Matsuyama (1992), Young (1991), Mountford (1998), and Baldwin, Philippe and Ottaviano (2001)] the focus on the interaction between population growth and comparative advantage as well as the persistent effect that this interaction may have on the distribution of population and income in the world economy, generates an important new insight regarding the distribution of the gains from trade. The theory suggests that even if trade affects output growth of the trading countries at the same rate (due to the terms of trade effect), income per capita of developed and less developed economies will diverge since in less developed economies growth of total output will be generated partly by population growth, whereas in developed economies it will be generated primarily by an increase in output per capita. 150 Consistent with the thesis that human capital has reinforced the existing patterns of comparative advantage, Taylor (1999) argues that human capital accumulation during the late 19th century was not a source of convergence even among the advanced ‘Greater Atlantic’ trading economies. The richer economies – the US and Australia – had greater levels of school enrollments than the poorer ones, Denmark and Sweden.
282
O. Galor
The historical evidence described in Section 2 suggests that the fundamental hypothesis of this theory is consistent with the process of development over the last two centuries. As implied by the trade patterns reported in Table 1, and the evolution of industrialization depicted in Figure 14, trade over this period induced the specialization of industrialized economies in the production of industrial goods, whereas non-industrial economies specialize in the production of primary goods. The asymmetric effect of international trade on the process of industrialization of developed and less developed economies, as depicted in Figure 14, affected the demand for human capital as analyzed in Section 2.3.3, and thus the timing of the demographic transition in developed and less-developed economies, generating a great divergence in output per capita as well as significant changes in the distribution of world population, as depicted in Figure 33.151 The diverging process of development of the UK and India since the 19th century in terms of the levels of income per capita and population growth is consistent with the theory of Galor and Mountford (2003) and provides an interesting case study. During the 19th century the UK traded manufactured goods for primary products with India.152 Trade with Asia constituted over 20% of UK total exports and 23.2% of total imports throughout the 19th century [Bairoch (1974)].153 Consistent with the proposed hypothesis, as documented in Figure 14, industrialization in the UK accelerated, leading to a significant increase in the demand for skilled labor in the second phase of the Industrial Revolution, a demographic transition and a transition to a state of sustained economic growth. For India, however, international trade played the reverse role. The period 1813–1850 was characterized by a rapid expansion in the volume of exports and imports which gradually transformed India from being an exporter of manufactured products – largely textiles – into a supplier of primary commodities [Chaudhuri (1983)]. Trade with the UK was fundamental in this process. The UK supplied over two thirds of its imports for most of the 19th century and was the market for over a third of India’s exports. As depicted in Figure 14, the rapid industrialization in the UK in the 19th century was associated with a decline in the per capita level of industrialization in India.154 The
151 Consistent with the viewpoint that trade has not been uniformly beneficial across time and regions, recent research by Rodriguez and Rodrik (2001) has indicated that the relationship between openness and growth changed in the last century. Moreover, Clemens and Williamson (2004) find a positive relationship between average tariff levels and growth for the period 1870–1913 and a negative relationship for the period 1970– 1998. Similarly Vamvakadis (2002) finds a positive relationship between several measures of openness and growth after 1970 and some evidence of a negative relationship in the period 1870–1910. 152 The colonial power of the UK may have encouraged the specialization of India in the production of primary goods beyond the degree dictated by market forces. However, these forces would have just reinforced the adverse effects described in their theory. 153 In contrast, trade with Asia constituted only 5% or less of French, German or Italian exports and 12.1% of total imports of continental Europe. 154 Furthermore, Bairoch (1974) found that industries that employed new technologies made up between 60% and 70% of the UK manufacturing industry in 1860 but less than 1% of manufacturing industries in the developing countries.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
283
setback in the process of industrialization and consequently the lack of demand for skilled labor, delayed the demographic transition and the process of development.155 Thus, while the gains from trade were utilized in the UK primarily towards an increase in output per capita, in India they were significantly channeled towards an increase in the size of the population. The ratio of output per capita in the UK relative to India grew from 3 : 1 in 1820 to 11 : 1 in 1998, whereas the ratio of India’s population relative to the UK’s population grew from 8 : 1 in 1870 to 16 : 1 in 1998.156
7. Concluding remarks The transition from stagnation to growth and the associated phenomenon of the great divergence have been the subject of an intensive research in the growth literature in recent years. The discrepancy between the predictions of exogenous and endogenous growth models and the process of development over most of human history, induced growth theorists to advance an alternative theory that would capture in a single unified framework the contemporary era of sustained economic growth, the epoch of Malthusian stagnation that had characterized most of human history, and the fundamental driving forces of the recent transition between these distinct regimes. The advancement of unified growth theory was fueled by the conviction that the understanding of the contemporary growth process would be limited and distorted unless growth theory would be based on micro-foundations that would reflect the qualitative aspects of the growth process in its entirety. In particular, the hurdles faced by less developed economies in reaching a state of sustained economic growth would remain obscured unless the origin of the transition of the currently developed economies into a state of sustained economic growth would be identified and its implications would 155 Unlike the rise in the industrial demand for education in the UK, education was not expanded to a similar degree in India in the 19th century. As noted by Basu (1974), during the 19th century the state of education in India was characterized by a relatively large university sector, aimed at producing skilled bureaucrats rather than industrialists, alongside widespread illiteracy of the masses. The literacy rate was very low, (e.g., 10% in Bengal in 1917–1918) but nevertheless, attempts to expand primary education in the 20th century were hampered by poor attendance and high drop out rates, suggesting that demand for education was relatively low. The lack of broad based education in India can also be seen using the data of Barro and Lee (2000). Despite an expansion of education throughout the 20th century Barro and Lee report that in 1960 72.2% of Indians aged 15 and above had “no schooling” compared with 2% in the UK. 156 Another interesting case study providing supporting evidence for the proposed hypothesis is the economic integration of the Israeli and the West Bank economies in the aftermath of the 1967 war. Trade and factor mobility between the skilled abundant economy of Israel and the unskilled abundant economy of the West Bank, shifted the West Bank economy towards further specialization in the production of primary goods, and possibly triggered the astonishing increase in crude birth rates from 22 per 1000 people in 1968 to 42 per 1000 in 1990, despite a decline in mortality rates. The gains from trade and development in the West Bank economy were converted primarily into an increase in population size, nearly doubling the population in those two decades. Estimates of the growth rates of output per capita over this period are less reliable and suggest that the increase was about 30%.
284
O. Galor
be modified to account for the additional economic forces faced by less developed economies in an interdependent world. Unified growth theory suggests that the transition from stagnation to growth is an inevitable outcome of the process of development. The inherent Malthusian interaction between the level of technology and the size and the composition of the population accelerated the pace of technological progress, and ultimately raised the importance of human capital in the production process. The rise in the demand for human capital in the second phase of industrialization and its impact on the formation of human capital, as well as on the onset of the demographic transition, brought about significant technological advancements along with a reduction in fertility rates and population growth, enabling economies to convert a larger share of the fruits of factor accumulation and technological progress into growth of income per capita, and paving the way for the emergence of sustained economic growth. Moreover, the theory suggests that differences in the timing of the take-off from stagnation to growth across countries contributed significantly to the Great Divergence and to the emergence of convergence clubs. Variations in the timing of the transition from stagnation to growth and thus in economic performance across countries (e.g., England’s earlier industrialization in comparison to China) reflect initial differences in geographical factors and historical accidents and their manifestation in variations in institutional, demographic, and cultural factors, trade patterns, colonial status, and public policy. In particular, once a technologically driven demand for human capital emerged in the second phase of industrialization, the prevalence of human capital promoting institutions determined the extensiveness of human capital formation, the timing of the demographic transition, and the pace of the transition from stagnation to growth. Thus, unified growth theory provides the natural framework of analysis in which variations in the economic performance across countries and regions could be examined based on the effect of variations in educational, institutional, geographical, and cultural factors on the pace of the transition from stagnation to growth. Further advancements of unified growth theory would necessitate refinements of some of the central building blocks of the theory as well as additional empirical and quantitative examinations of the fundamental hypothesis based on contemporary and historical data. In particular, while the micro foundations for fertility decisions, population growth, and to a lesser extent human capital formation, appears profound in existing unified theories, the modeling of the factors that govern technological progress could be enhanced using recent insights from the theory of endogenous technological change as well as from the recent advancements in the study of the role of human capital and institutional factors in technological progress. Moreover, unified growth theory provides a new set of testable predictions that could guide economic historians in their data collection, as well as in revising their past interpretations of existing historical evidence, enhancing the refinements of the main hypotheses of unified growth theory. The most promising and challenging future research in the field of economic growth in the next decades would be the exploration of the interaction between human evolution and the process of economic development. This research will revolutionize our
Ch. 4: From Stagnation to Growth: Unified Growth Theory
285
understanding of the process of economic development as well as the process of human evolution, establishing socio-biological evolutionary foundations to the growth process.
Acknowledgements The author wishes to thank Philippe Aghion, Graziella Bertocchi, Carl Johan Dalgaard, Matthias Doepke, Hagai Etkes, Moshe Hazan, Nils-Petter Lagerlof, Sebnem KalemliOzcan, Daniel Mejia, Joel Mokyr, Omer Moav, Andrew Mountford, Nathan Sussman, and David Weil for valuable discussions and detailed comments, and Tamar Roth for excellent research assistance. This research is supported by a NSF Grant SES-0004304.
References Abramovitz, M., David, P.A. (2000). “American macroeconomic growth in the era of knowledge-based progress: The long-run perspective”. In: Engerman, S.L., Gallman, R.E. (Eds.), The Cambridge Economic History of the United States, vol. 2. Cambridge University Press, New York. Acemoglu, D., Aghion, P., Zilibotti, F. (2004). Distance to Frontier, Selection, and Economic Growth. MIT Press, Cambridge, MA. Acemoglu, D., Johnson, S., Robinson, J.A. (2002). “Reversal of fortune: Geography and institutions in the making of the modern world income distribution”. Quarterly Journal of Economics 117, 1231–1294. Acemoglu, D., Zilibotti, F. (1997). “Was Prometheus unbound by chance? Risk, diversification, and growth”. Journal of Political Economy 105, 709–751. Aghion, P., Howitt, P. (1992). “A model of growth through creative destruction”. Econometrica 60, 323–351. Aghion, P., Howitt, P., Mayer-Foulkes, D. (2005). “The effect of financial development on convergence: Theory and evidence”. Quarterly Journal of Economics 120, 173–222. Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S., Wacziarg, R. (2003). “Fractionalization”. Journal of Economic Growth 8, 155–194. Anderson, R.D. (1975). Education in France 1848–1870. Clarendon Press, Oxford. Andorka, R. (1978). Determinants of Fertility in Advanced Societies. Free Press, New York. Azariadis, C. (1996). “The economics of poverty traps”. Journal of Economic Growth 1, 449–486. Baldwin, R.E., Philippe, M., Ottaviano, G.I.P. (2001). “Global income divergence, trade and industrialization: The geography of growth take-offs”. Journal of Economic Growth 6, 5–37. Bairoch, P. (1974). “Geographical structure and trade balance of European foreign trade from 1800–1970”. Journal of European Economic History 3, 557–608. Bairoch, P. (1982). “International industrialization levels from 1750–1980”. Journal of European Economic History 11, 269–333. Bairoch, P. (1988). Cities and Economic Development. University of Chicago Press, Chicago. Barro, R.J., Becker, G.S. (1989). “Fertility choice in a model of economic growth”. Econometrica 57, 481– 501. Barro, R.J., Lee, J. (2000). “International data on educational attainment: Updates and implications”. Harvard University. Barro, R.J., Sala-i-Martin, X. (2003). Economic Growth. MIT Press, Cambridge, MA. Basu, A. (1974). The Growth of Education and Political Development in India 1898–1920. Oxford University Press, Oxford. Becker, G.S. (1981). A Treatise on the Family. Harvard University Press, Cambridge, MA. Becker, G.S., Lewis, H.G. (1973). “On the interaction between the quantity and quality of children”. Journal of Political Economy 81, S279–S288.
286
O. Galor
Becker, G.S., Murphy, K., Tamura, R. (1990). “Human capital, fertility, and economic growth”. Journal of Political Economy 98, S12–S37. Benabou, R. (2000). “Unequal societies: Income distribution and the social contract”. American Economic Review 90, 96–129. Ben Porath, Y. (1967). “The production of human capital and the life cycle of earnings”. Journal of Political Economy 75, 352–365. Berdugo, B., Sadik, J., Sussman, N. (2003). “Delays in technology adoption, appropriate human capital, natural resources and growth”. Hebrew University. Berghahn, V.R. (1994). Imperial Germany, 1871–1914: Economy, Society, Culture and Politics. Berghahn Books, Providence, RI. Bertocchi, G. (2003). “The law of primogeniture and the transition from landed aristocracy to industrial democracy”. CEPR Discussion Paper 3723. Bertocchi, G., Canova, F. (2002). “Did colonization matter for growth? An empirical exploration into the historical causes of Africa’s underdevelopment”. European Economic Review 46, 1851–1871. Bertocchi, G., Spagat, M. (2004). “The evolution of modern educational systems: Technical vs. general education, distributional conflict, and growth”. Journal of Development Economics 73, 559–582. Bisin, A., Verdier, T. (2000). “Beyond the melting pot: Cultural transmission, marriage, and the evolution of ethnic and religious traits”. Quarterly Journal of Economics 115, 955–988. Bloom, D.E., Canning, D., Sevilla, J. (2003). “Geography and poverty traps”. Journal of Economic Growth 4, 355–378. Bloom, D., Williamson, J.G. (1998). “Demographic transition and economic miracles in emerging Asia”. World Bank Economic Review 12, 419–455. Boldrin, M., Jones, L. (2002). “Mortality, fertility, and saving in a Malthusian economy”. Review of Economic Dynamics 5, 775–814. Borghans, J.A.M., Borghans, L., ter-Weel, B. (2004). “Economic performance, human cooperation, and the major histocompatibility complex”. Maastricht. Boserup, E. (1965). The Conditions of Agricultural Progress. Aldine Publishing Company, Chicago. Boucekkine, R., de la Croix, D., Licandro, O. (2003). “Early mortality declines at the dawn of modern growth”. Scandinavian Journal of Economics 105, 401–418. Bourguignon, F., Verdier, T. (2000). “Oligarchy, democracy, inequality and growth”. Journal of Development Economics 62, 285–313. Bowles, S. (1998). “Endogenous preferences: The cultural consequences of markets and other economic institutions”. Journal of Economic Literature 36, 75–111. Bowles, S., Gintis, H. (1975). “Capitalism and education in the United States”. Socialist Revolution 5, 101– 138. Boyd, R., Richardson, P.J. (1985). Culture and the Evolutionary Process. University of Chicago Press, Chicago. Boyer, G. (1989). “Malthus was right after all: Poor relief and birth rates in South-Eastern England”. Journal of Political Economy 97, 93–114. Brezis, E.S., Krugman, P.R., Tsiddon, D. (1993). “Leapfrogging in international competition: A theory of cycles in national technological leadership”. American Economic Review 83, 1211–1219. Browning, M., Hansen, L.P., Heckman, J.J. (1999). “Micro data and general equilibrium models”. In: Taylor, J., Woodford, M. (Eds.), Handbook of Macroeconomics. North-Holland, Amsterdam. Caldwell, W.J. (1976). “Toward a restatement of demographic transition theory”. Population and Development Review 2, 321–366. Cameron, R. (1989). A History of Western Education. Vol. 3: The Modern Europe and the New World. Oxford University Press, Oxford. Caselli, F., Coleman, W.J. (2002). “The world technological frontier”. Harvard University. Cavalli-Sforza, L.L., Feldman, M.W. (1981). Cultural Transmission and Evolution: A Quantitative Approach. Princeton University Press, Princeton. Cervellati, M., Sunde, U. (2005). “Human capital formation, life expectancy and the process of development”. American Economic Review. In press.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
287
Chanda, A., Dalgaard, C.-J. (2003). “Dual economies and international total factor productivity differences”. University of Copenhagen. Chaudhuri, K.N. (1983). “Foreign trade and balance of payments (1757–1947)”. In: Kumar, D. (Ed.), The Cambridge Economic History of India. Cambridge University Press, Cambridge, MA. Chesnais, J. (1992). The Demographic Transition: Stages, Patterns and Economic Implications. Clarendon Press, Oxford. Cipolla, C.M. (1969). Literacy and Development in the West. Penguin Books, Middlesex. Clark, G. (2001). “The secret history of the Industrial Revolution”. UC Davis. Clark, G. (2002). “Farmland rental values and agrarian history: England and Wales, 1500–1912”. European Review of Economic History 6, 281–308. Clark, G. (2003). “The condition of the working-class in England, 1200–2000: Magna Carta to Tony Blair”. UC Davis. Clark, G., Hamilton, G. (2003). “Survival of the fittest? Capital, human capital, and reproduction in European Society before the Industrial Revolution”. UC Davis. Clemens, M.A., Williamson, J.G. (2004). “Why did the tariff-growth correlation reverse after 1950?”. Journal of Economic Growth 9, 5–46. Coale, A.J., Treadway, R. (1986). “A summary of the changing distribution of overall fertility, marital fertility, and the proportion married in the provinces of Europe”. In: Coale, A.J., Watkins, S. (Eds.), The Decline of Fertility in Europe. Princeton University Press, Princeton. Cody, M.L. (1966). “A general theory of clutch size”. Evolution 20, 174–184. Cohen, M.N. (1989). Health and the rise of civilization. Yale University Press, New Haven. Connolly, M., Peretto, P.F. (2003). “Industry and the family: Two engines of growth”. Journal of Economic Growth 8, 115–148. Crafts, N.F.R. (1985). British Economic Growth during the Industrial Revolution. Oxford University Press, Oxford. Crafts, N.F.R., Harley, C.K. (1992). “Output growth and the Industrial Revolution: A restatement of the Crafts–Harley view”. Economic History Review 45, 703–730. Craig, F.W.S. (1989). British Electoral Facts, 1832–1987. Gower Press, Brookfield. Cressy, D. (1980). Literacy and the Social Order: Reading and Writing in Tudor and Stuart England. Cambridge University Press, Cambridge, MA. Cressy, D. (1981). “Levels of illiteracy in England 1530–1730”. In: Graff, H.J. (Ed.), Literacy and Social Development in the West: A Reader. Cambridge University Press, Cambridge, MA. Cubberly, E.P. (1920). The History of Education. Cambridge University Press, Cambridge, MA. Dahan, M., Tsiddon, D. (1998). “Demographic transition, income distribution, and economic growth”. Journal of Economic Growth 3, 29–52. Dalgaard, C.-J., Kreiner, C.T. (2001). “Is declining productivity inevitable?”. Journal of Economic Growth 6, 187–204. Darwin, C. (1859). On the Origin of Species by Means of Natural Selection. John Murray, London. Darwin, C. (1871). The Descent of Man, and Selection in Relation to Sex. John Murray, London. Dawkins, R. (1989). The Selfish Gene. Oxford University Press, Oxford. De la Croix, D., Doepke, M. (2003). “Inequality and growth: Why differential fertility matters”. American Economic Review 93, 1091–1113. De Vries, J. (1984). European Urbanization, 1500–1800. Harvard University Press, Cambridge, MA. Diamond, J. (1997). Guns, Germs, and Steel: The Fates of Human Societies. Norton, New York. Doepke, M. (2004). “Accounting for fertility decline during the transition to growth”. Journal of Economic Growth 9, 347–383. Doepke, M. (2005). “Child mortality and fertility decline: Does the Barro–Becker model fit the facts?”. Journal of Population Economics 17, 337–366. Doepke, M., Zilibotti, F. (2005). “The macroeconomics of child labor regulation”. American Economic Review 95. Doms, M., Dunne, T., Troske, K.R. (1997). “Workers, wages and technology”. Quarterly Journal of Economics 112, 253–290.
288
O. Galor
Duffy, J., Papageorgiou, C., Perez-Sebastian, F. (2004). “Capital-skill complementarity? Evidence from a panel of countries”. Review of Economic and Statistics 86, 327–344. Durham, W. (1982). “Interaction of genetic and cultural evolution: Models and examples”. Human Ecology 10, 289–323. Durlauf, S.N. (1996). “A theory of persistent income inequality”. Journal of Economic Growth 1, 75–94. Durlauf, S.N., Johnson, P.A. (1995). “Multiple regimes and cross-country growth behavior”. Journal of Applied Econometrics 10, 365–384. Durlauf, S.N., Quah, D. (1999). “The new empirics of economic growth”. In: Taylor, J.B., Woodford, M. (Eds.), Handbook of Macroeconomics. North-Holland, Amsterdam. Dyson, T., Murphy, M. (1985). “The onset of fertility transition”. Population and Development Review 11, 399–440. Eckstein, Z., Mira, P., Wolpin, K.I. (1999). “A quantitative analysis of Swedish fertility dynamics: 1751– 1990”. Review of Economic Dynamics 2, 137–165. Easterlin, R. (1981). “Why isn’t the whole world developed?”. Journal of Economic History 41, 1–19. Easterly, W., Levine, R. (1997). “Africa’s growth tragedy: Policies and ethnic divisions”. Quarterly Journal of Economics 111, 1203–1250. Easterly, W., Levine, R. (2003). “Tropics, germs, and crops: the role of endowments in economic development”. Journal of Monetary Economics 50, 3–39. Edlund, L., Lagerlof, N.-P. (2002). “Implications of marriage institutions for redistribution and growth”. Columbia University. Endler, J.A. (1986). Natural Selection in the Wild. Princeton University Press, Princeton. Engerman, S., Sokoloff, K.L. (2000). “Factor endowment, inequality, and paths of development among New World economies”. UCLA. Erlich, I., Lui, F.T. (1991). “Intergenerational trade, longevity, and economic growth”. Journal of Political Economy 99, 1029–1059. Estevadeordal, A., Frantz, B., Taylor, A.M. (2002). “The rise and fall of world trade, 1870–1939”. Quarterly Journal of Economics 118, 359–407. Feinstein, C.H. (1972). National Income, Expenditure and Output of the United Kingdom 1855–1965. Cambridge University Press, Cambridge, MA. Fernandez, R., Fogli, A., Olivetti, C. (2004). “Mothers and sons: Preference formation and female labor force dynamics”. Quarterly Journal of Economics 119, 1249–1301. Fernandez, R., Rogerson, R. (1996). “Income distribution, communities, and the quality of public education”. Quarterly Journal of Economics 111, 135–164. Fernandez-Villaverde, J. (2005). “Was Malthus right? Economic growth and population dynamics”. University of Pennsylvania. Feyrer, J. (2003). “Convergence by parts”. Dartmouth College. Fiaschi, D., Lavezzi, A.M. (2003). “Distribution dynamics and nonlinear growth”. Journal of Economic Growth 8, 379–402. Field, A. (1976). “Educational reform and manufacturing development in mid-nineteenth century Massachusetts”. Journal of Economic History 36, 263–266. Findlay, R., Keirzkowsky, H. (1983). “International trade and human capital: A simple general equilibrium model”. Journal of Political Economy 91, 957–978. Findlay, R., O’Rourke, K.H. (2003). “Commodity market integration, 1500–2000”. In: Bordo, M.D., Taylor, A.M., Williamson, J.G. (Eds.), Globalization in Historical Perspective. University of Chicago Press, Chicago. Flora, P., Kraus, F., Pfenning, W. (1983). State, Economy and Society in Western Europe 1815–1975, vol. 1. St. James Press, Chicago. Foster, A.D., Rosenzweig, M.R. (1996). “Technical change and human-capital returns and investments: Evidence from the Green Revolution”. American Economic Review 86, 931–953. Galor, O. (1996). “Convergence?: Inferences from theoretical models”. Economic Journal 106, 1056–1069. Galor, O. (2005). “The demographic transition and the emergence of sustained economic growth”. Journal of the European Economic Association 3, 494–504.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
289
Galor, O., Moav, O. (2000). “Ability biased technological transition, wage inequality and growth”. Quarterly Journal of Economics 115, 469–498. Galor, O., Moav, O. (2002). “Natural selection and the origin of economic growth”. Quarterly Journal of Economics 117, 1133–1192. Galor, O., Moav, O. (2006). “Das human Kapital: A theory of the demise of the class structure”. Review of Economic Studies 73. Galor, O., Moav, O. (2004a). “From physical to human capital accumulation: Inequality and the process of development”. Review of Economic Studies 71, 1001–1026. Galor, O., Moav, O. (2004b). “Natural selection and the evolution of life expectancy”, Brown University. Galor, O., Moav, O., Vollrath, D. (2003). “Land inequality and the origin of divergence and overtaking in the growth process: Theory and evidence”. Brown University. Galor, O., Mountford, A. (2003). “Trading population for productivity”. Brown University. Galor, O., Tsiddon, D. (1997). “Technological progress, mobility, and growth”. American Economic Review 87, 363–382. Galor, O., Weil, D.N. (1996). “The gender gap, fertility, and growth”. American Economic Review 86, 374– 387. Galor, O., Weil, D.N. (1999). “From Malthusian stagnation to modern growth”. American Economic Review 89, 150–154. Galor, O., Weil, D.N. (2000). “Population, technology and growth: From the Malthusian regime to the demographic transition”. American Economic Review 110, 806–828. Galor, O., Zeira, J. (1993). “Income distribution and macroeconomics”. Review of Economic Studies 60, 35–52. Gallup, J.L., Sachs, J.D., Mellinger, A.D. (1998). “Geography and economic development”. NBER Working Paper No. w6849. Glaeser, E.L., La Porta, R., Lopez-de-Silanes, F., Shleifer, A. (2004). “Do institutions cause growth?”. Journal of Economic Growth 9, 271–303. Goldin, C. (1990). Understanding The Gender Gap: An Economic History of American Women. Oxford University Press, New York. Goldin, C. (2001). “The human capital century and American leadership: Virtues of the past”. Journal of Economic History 61, 263–292. Goldin, C., Katz, L.F. (1998). “The origins of technology-skill complementarity”. Quarterly Journal of Economics 113, 693–732. Goldin, C., Katz, L.F. (2001). “On the legacy of U.S. educational leadership: Notes on distribution and economic growth in the 20th century”. American Economic Review 91, 18–23. Goodfriend, M., McDermott, J. (1995). “Early development”. American Economic Review 85, 116–133. Gould, E.D., Moav, O., Simhon, A. (2003). “The mystery of monogamy”. Hebrew University. Graham, B.S., Temple, J.R.W. (2004). “Rich nations, poor nations: how much can multiple equilibria explain?”. University of Bristol. Grant, B.R., Grant, P.R. (1989). Evolutionary Dynamics of a Natural Population. University of Chicago Press, Chicago. Green, A. (1990). Education and State Formation. St. Martin’s Press, New York. Greenwood, J., Seshadri, A. (2002). “The U.S. demographic transition”. American Economic Review 92, 153–159. Greenwood, J., Seshadri, A., Yorukoglu, M. (2005). “Engines of liberation”. Review of Economic Studies 72, 109–133. Grossman, G.M., Helpman, E. (1991). Innovation and Growth. MIT Press, Cambridge. Grossman, H.I., Kim, M. (1999). “Education policy: Egalitarian or elitist?”. Economics and Politics 15, 225– 246. Hall, R.E., Jones, C.I. (1999). “Why do some countries produce so much more output per worker than others?”. Quarterly Journal of Economics 114, 83–116. Hansen, G., Prescott, E. (2002). “Malthus to Solow”. American Economic Review 92, 1205–1217.
290
O. Galor
Hanushek, E.A. (1992). “The trade-off between child quantity and quality”. Journal of Political Economy 100, 84–117. Hassler, J., Rodriguez Mora, J.V. (2000). “Intelligence, social mobility, and growth”. American Economic Review 90, 888–908. Hazan, M., Berdugo, B. (2002). “Child labor, fertility and economic growth”. Economic Journal 112, 810– 828. Hazan, M., Zoabi, H. (2004). “Does longevity cause growth?”. Hebrew University. Hernandez, D.J. (2000). Trends in the Well Being of America’s Children and Youth. U.S. Bureau of the Census, Washington, DC. Hibbs, D.A., Olson, O. (2005). “Biogeography and long-run economic development”. European Economic Review 48. Horrell, S., Humphries, J. (1995). “The exploitation of little children: Child labor and the family economy in the industrial revolution”. Exploration in Economic History 32, 485–516. Howitt, P., Mayer-Foulkes, D. (2005). “R&D, implementation and stagnation: A Schumpeterian theory of convergence clubs”. Journal of Money Credit and Banking 37. Human Mortality Database (2003). University of California, Berkeley, USA, and Max Planck Institute for Demographic Research, Germany. Hurt, J. (1971). Education in Evolution. Paladin, London. Iyigun, M.F. (2005). “Geography, demography, and early development”. Journal of Population Economics 17. Jones, C.I. (1997). “Convergence revisited”. Journal of Economic Growth 2, 131–154. Jones, C.I. (2001). “Was an Industrial Revolution inevitable? Economic growth over the very long run”. Advances in Macroeconomics 1, 1–43. Jones, E.L. (1981). The European Miracle: Environments, Economies and Geopolitics in the History of Europe and Asia. Cambridge University Press, Cambridge, CA. Kalemli-Ozcan, S. (2002). “Does the mortality decline promote economic growth”. Journal of Economic Growth 7, 411–439. Kettlewell, H.B.D. (1973). The Evolution of Melanism. Clarendon Press, Oxford. Kirkwood, T.B.L. (1998). “Evolution theory and the mechanisms of aging”. In: Brocklehurst, J.C., Tallis, R.C. (Eds.), Textbook of Geriatric Medicine. 5th ed. Churchill Livingstone, London, pp. 45–49. Kogel, T., Prskawetz, A. (2001). “Agricultural productivity growth and escape from Malthusian trap”. Journal of Economic Growth 6, 337–357. Kohler, H., Rodgers, J.L., Christensen, K. (1999). “Is fertility behavior in our genes? Findings from a Danish twin study”. Population and Development Review 25, 253–263. Komlos, J., Artzrouni, M. (1990). “Mathematical investigations of the escape from the Malthusian trap”. Mathematical Population Studies 2, 269–287. Kremer, M. (1993). “Population growth and technological change: One million B.C. to 1990”. Quarterly Journal of Economics 108, 681–716. Kremer, M., Chen, D.L. (2002). “Income distribution dynamics with endogenous fertility”. Journal of Economic Growth 7, 227–258. Kuhn, T.S. (1957). The Copernican Revolution. Harvard University Press, Cambridge, MA. Kurian, G.T. (1994). Datapedia of the U.S. 1790–2000, America Year by Year. Bernan Press, Lonham. Kuznets, S. (1967). “Quantitative aspects of the economic growth of nations: X-level and structure of foreign trade: Long-term trends”. Economic Development and Cultural Change 15, 1–140. Kuzynski, R.R. (1969). The Measurement of Population Growth. Gordon and Breach Science Publishers, New York. Lack, D. (1954). The Natural Regulation of Animal Numbers. Clarendon Press, Oxford. Lagerlof, N. (2003a). “From Malthus to modern growth: The three regimes revisited”. International Economic Review 44, 755–777. Lagerlof, N. (2003b). “Gender equality and long-run growth”. Journal of Economic Growth 8, 403–426. Lagerlof, N. (2006). “The Galor–Weil model revisited: A quantitative exploration”. Review of Economic Dynamics 9.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
291
Landes, D.S. (1969). The Unbound Prometheus. Technological Change and Industrial Development in Western Europe from 1750 to the Present. Cambridge University Press, Cambridge, MA. Landes, D.S. (1998). The Wealth and Poverty of Nations. Norton, New York. Lee, R.D. (1997). “Population dynamics: Equilibrium, disequilibrium, and consequences of fluctuations”. In: Stark, O., Rosenzweig, M. (Eds.), The Handbook of Population and Family Economics. Elsevier, Amsterdam. Levy-Leboyer, M., Bourguignon, F. (1990). The French Economy in the Nineteenth Century. Cambridge University Press, Cambridge, MA. Lindert, P.H., Williamson, J.G. (1976). “Three centuries of American inequality”. Research in Economic History 1, 69–123. Livi-Bacci, M. (1997). A Concise History of World Population. Blackwel, Oxford. Livingston, F. (1958). “Anthropological implications of sickle cell distribution in West Africa”. American Anthropologist 60, 533–562. Lucas, R.E. (2002). The Industrial Revolution: Past and Future. Harvard University Press, Cambridge, MA. MacArthur, R.H., Wilson, E.O. (1967). The Theory of Island Biogeography. Princeton University Press, Princeton. Maddison, A. (2001). The World Economy: A Millennia Perspective. OECD, Paris. Maddison, A. (2003). The World Economy: Historical Statistics. CD-ROM. OECD, Paris. Malthus, T.R. (1798). An Essay on the Principle of Population. St. Paul’s Church-Yard, London. Masters, W.E., McMillan, M.S. (2001). “Climate and scale in economic growth”. Journal of Economic Growth 6, 167–187. Matsuyama, K. (1992). “Agricultural productivity, comparative advantage, and economic growth”. Journal of Economic Theory 58, 317–334. Matthews, R.C., Feinstein, C.H., Odling-Smee, J.C. (1982). British Economic Growth 1856–1973. Stanford University Press, Stanford. McClelland, C.E. (1980). State, Society, and University in Germany: 1700–1914. Cambridge University Press, Cambridge, MA. McDermott, J. (2002). “Development dynamics: Economic integration and the demographic transition”. Journal of Economic Growth 7, 371–410. Mitch, D. (1992). The Rise of Popular Literacy in Victorian England: The Influence of Private Choice and Public Policy. University of Pennsylvania Press, Philadelphia, PA. Mitch, D. (1993). “The role of human capital in the First Industrial Revolution”. In: Mokyr, J. (Ed.), The British Industrial Revolution: An Economic Perspective. Westview Press, Boulder. Mitchell, B. (1981). European Historical Statistics, 1750–1975, second ed. New York University Press, New York. Moav, O. (2005). “Cheap children and the persistence of poverty”. Economic Journal 115, 88–110. Mokyr, J. (1985). Why Ireland Starved: A Quantitative and Analytical History of the Irish Economy, 1800– 1850. Allen and Unwin, London. Mokyr, J. (1990). The Lever of Riches. Oxford University Press, New York. Mokyr, J. (1993). “The new economic history and the Industrial Revolution”. In: Mokyr, J. (Ed.), The British Industrial Revolution: An Economic Perspective. Westview Press, Boulder. Mokyr, J. (2001). “The rise and fall of the factory system: Technology, firms, and households since the Industrial Revolution”. Carnegie–Rochester Conference Series on Public Policy 55, 1–45. Mokyr, J. (2002). The Gifts of Athena: Historical Origins of the Knowledge Economy. Princeton University Press, Princeton. Mookherjee, D., Ray, D. (2003). “Persistent inequality”. Review of Economic Studies 70, 369–393. Morrisson, C., Snyder, W. (2000). “Income inequalities in France from the early eighteenth century to 1985”. Revue Economique 51, 119–154. Mountford, A. (1998). “Trade, convergence and overtaking”. Journal of International Economics 46, 167–182. Muller, D.K. (1987). “The process of systematization: The case of German secondary education”. In: Muller, D., Ringer, F., Simon, B. (Eds.), The Rise of the Modern Educational System. Cambridge University Press, Cambridge, MA.
292
O. Galor
Neher, A.P. (1971). “Peasants, procreation and pensions”. American Economic Review 61, 380–389. Nelson, R.R., Phelps, E.S. (1966). “Investment in humans, technological diffusion, and economic growth”. American Economic Review 56, 69–75. North, D.C. (1981). Structure and Change in Economic History. Norton, New York. Ofek, H. (2001). Second Nature: Economic Origin of Human Evolution. Cambridge University Press, Cambridge, MA. O’Rourke, K.H., Williamson, J.G. (1999). Globalization and History. MIT Press, Cambridge, MA. O’Rourke, K.H., Williamson, J.G. (2005). “Malthus to Ohlin”. Journal of Economic Growth 10. Parente, S., Prescott, E.C. (2000). Barriers to Riches. MIT Press, Cambridge, MA. Pereira, A.S. (2003). “When did modern economic growth really start? The empirics of Malthus to Solow”. UBC. Pomeranz, K. (2000). The Great Divergence: China, Europe and the Making of the Modern World Economy. Princeton University Press, Princeton. Pritchett, L. (1997). “Divergence, big time”. Journal of Economic Perspectives 11, 3–17. Przeworski, A. (2003). “The last instance: Are institutions the primary cause of economic development?”. Department of Politics, New York University. Psacharopoulos, G., Patrinos, H.A. (2002). “Returns to investment in education: A further update”. The World Bank. Quah, D. (1996). “Convergence empirics across economies with (some) capital mobility”. Journal of Economic Growth 1, 95–124. Quah, D. (1997). “Empirics for growth and distribution: Stratification, polarization, and convergence clubs”. Journal of Economic Growth 2, 27–61. Razin, A., Ben-Zion, U. (1975). “An intergenerational model of population growth”. American Economic Review 65, 923–933. Ringer, F. (1979). Education and Society in Modern Europe. Indiana University Press, Bloomington. Robson, A.J. (2001). “The biological basis of economic behavior”. Journal of Economic Literature 39, 11–33. Robson, A.J., Kaplan, H.S. (2003). “The evolution of human longevity and intelligence in Hunter–Gatherer economies”. American Economic Review 93, 150–169. Rodgers, J.L., Doughty, D. (2000). “Genetic and environmental influences on fertility expectations and outcomes using NLSY kinship data”. In: Rodgers, J.L., Rowe, D.C., Miller, W.B. (Eds.), Genetic Influences on Human Fertility and Sexuality. Kluwer, Boston. Rodgers, J.L., Hughes, K., Kohler, H., Christensen, K., Doughty, D., Rowe, D.C., Miller, W.B. (2001a). “Genetic influence helps explain variation in human fertility: Evidence from recent behavioral and molecular genetic studies”. Current Directions in Psychological Science 10, 184–188. Rodgers, J.L., Kohler, H., Ohm Kyvik, K., Christensen, K. (2001b). “Behavior genetic modeling of human fertility: Findings from a contemporary Danish twin study”. Demography 38, 29–42. Rodriguez, R., Rodrik, D. (2001). “Trade policy and economic growth: A skeptic’s guide to the cross-national evidence”. In: Bernanke, B., Rogoff, K.S. (Eds.), NBER Macroeconomics Annual. MIT Press, Cambridge, MA. Rodrik, D., Subramanian, A., Trebbi, F. (2004). “Institutions rule: The primacy of institutions over geography and integration in economic development”. Journal of Economic Growth 9, 131–165. Romer, P.M. (1990). “Endogenous technological change”. Journal of Political Economy 98, S71–S102. Rosenzweig, M.R., Wolpin, K.I. (1980). “Testing the quantity-quality fertility model: The use of twins as a natural experiment”. Econometrica 48, 227–240. Saint-Paul, G. (2003). “On market and human evolution”. CEPR Discussion Paper No. 3654. Sala-i-Martin, X. (2002). “The disturbing “rise” in world income distribution”. Columbia University. Sanderson, M. (1995). Education, Economic Change and Society in England 1780–1870. Cambridge University Press, Cambridge, MA. Schofield, R.S. (1973). “Dimensions of illiteracy, 1750–1850”. Explorations in Economic History 10, 437– 454. Schultz, T.W. (1964). Transforming Traditional Agriculture. Yale University Press, New Haven.
Ch. 4: From Stagnation to Growth: Unified Growth Theory
293
Schultz, T.W. (1975). “The value of the ability to deal with disequilibria”. Journal of Economic Literature 8, 827–846. Simon, B. (1987). “Systematization and segmentation in education: The case of England”. In: Muller, D., Ringer, F., Simon, B. (Eds.), The Rise of the Modern Educational System. Cambridge University Press, Cambridge, MA. Smith, A. (1776). The Wealth of Nations. Modern Library, New York. Soares, R.R. (2005). “Mortality reductions, educational attainment, and fertility choice”. American Economic Review 95 (3). Solow, R.M. (1956). “A contribution to the theory of economic growth”. Quarterly Journal of Economics 70, 65–95. Spree, R. (1977). Die Wachstumszyklen der deutschen Wirtschaft von 1840 bis 1880. Dunker & Humboult, Berlin. Stokey, N. (2001). “A quantitative model of the British Industrial Revolution, 1780–1850”. Carnegie– Rochester Conference Series on Public Policy 55, 55–109. Stone, L. (1969). “Literacy and education in England 1640–1900”. Past and Present 42, 69–139. Tamura, R.F. (2002). “Human capital and the switch from agriculture to industry”. Journal of Economic Dynamics and Control 27, 207–242. Tamura, R.F. (2004). “Human capital and economic development”. Clemson. Taylor, A.M. (1999). “Sources of convergence in the late nineteenth century”. European Economic Review 9, 1621–1645. U.S. Bureau of the Census (1975). Historical Statistics of the United States: Colonial Times to 1970. U.S. Bureau of the Census, Washington, DC. Vamvakidis, A. (2002). “How robust is the growth–openness connection? Historical evidence”. Journal of Economic Growth 7, 57–80. Voth, H.-J. (2003). “Living standards during the Industrial Revolution: An economist’s guide”. American Economic Review 93, 221–226. Voth, H.-J. (2004). “Living standards and the urban environment”. In: Johnson, P., Floud, R. (Eds.), The Cambridge Economic History of England. Cambridge University Press, Cambridge, MA. Weil, D.N. (2004). Economic Growth. Addison-Wesley, Boston, MA. Weisdorf, J.L. (2004). “From stagnation to growth: Revisiting three historical regimes”. Journal of Population Economics 17, 455–472. Weisenfeld, S.L. (1975). “Sickle cell trait in human biological and cultural evolution”. Science 157, 1135– 1140. Williamson, J.G. (1985). “Did British Capitalism Breed Inequality?”. Allen & Unwin, Boston. Wolthuis, J. (1999). “Lower technical education in the Netherlands 1798–1993: The rise and fall of a subsystem”. Ph.D. Thesis. Rijksuniversiteit Groningen, Netherlands. World Development Indicators. (2001). The World Bank, Washington, DC. Wrigley, E.A. (1969). Population and History. McGraw-Hill, New York. Wrigley, E.A., Schofield, R.S. (1981). The Population History of England 1541–1871: A Reconstruction. Harvard University Press, Cambridge, MA. Yates, P.L. (1959). Forty Years of Foreign Trade: A Statistical Handbook with Special Reference to Primary Products and Underdeveloped Countries. Allen & Unwin, London. Young, A. (1991). “Learning by doing and the dynamic effects of international trade”. Quarterly Journal of Economics 106, 369–405.
Chapter 5
POVERTY TRAPS* COSTAS AZARIADIS Department of Economics, University of California Los Angeles, 405 Hilgard Avenue, Los Angeles, CA 90095-1477, USA e-mail:
[email protected] JOHN STACHURSKI Department of Economics, The University of Melbourne, VIC 3010, Australia e-mail:
[email protected]
Contents Abstract Keywords 1. Introduction 2. Development facts 2.1. Poverty and riches 2.2. A brief history of economic development
3. Models and definitions 3.1. Neoclassical growth with diminishing returns 3.2. Convex neoclassical growth and the data 3.3. Poverty traps: historical self-reinforcement 3.4. Poverty traps: inertial self-reinforcement
4. Empirics of poverty traps 4.1. Bimodality and convergence clubs 4.2. Testing for existence 4.3. Model calibration 4.4. Microeconomic data
5. Nonconvexities, complementarities and imperfect competition 5.1. Increasing returns and imperfect competition 5.2. The financial sector and coordination 5.3. Matching
296 296 297 303 303 304 307 307 312 317 326 330 330 335 337 339 340 341 343 346
* This chapter draws on material contained in two earlier surveys by the first author [Azariadis, C. (1996). “The economics of poverty traps”. Journal of Economic Growth 1, 449–486; Azariadis, C. (2005). “The theory of poverty traps: what have we learned?”. In: Bowles S., Durlauf S., Hoff, K. (Eds.), Poverty Traps. Princeton University Press, Princeton].
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01005-1
296
C. Azariadis and J. Stachurski 5.4. Other studies of increasing returns
6. Credit markets, insurance and risk 6.1. Credit markets and human capital 6.2. Risk 6.3. Credit constraints and endogenous inequality
7. Institutions and organizations 7.1. Corruption and rent-seeking 7.2. Kinship systems
8. Other mechanisms 9. Conclusions 9.1. Lessons for economic policy
Acknowledgements Appendix A A.1. Markov chains and ergodicity A.2. Remaining proofs
References
349 350 351 355 358 363 364 367 373 373 374 375 375 375 378 379
Abstract This survey reviews models of self-reinforcing mechanisms that cause poverty to persist. Some of them examine market failure in environments where the neoclassical assumptions on markets and technology break down. Other mechanisms include institutional failure which can, by itself, perpetuate self-reinforcing poverty. A common thread in all these mechanisms is their adverse impact on the acquisition of physical or human capital, and on the adoption of modern technology. The survey also reviews recent progress in the empirical poverty trap literature.
Keywords world income distribution, persistent poverty, market failure, institutions, history dependence, technology JEL classification: 011, 012, 040
Ch. 5: Poverty Traps
297
In the problem of economic development, a phrase that crops up frequently is ‘the vicious circle of poverty’. It is generally treated as something obvious, too obvious to be worth examining. I hope I may be forgiven if I begin by taking a look at this obvious concept. [R. Nurkse (1953)]
1. Introduction Despite the considerable amount of research devoted to economic growth and development, economists have not yet discovered how to make poor countries rich. As a result, poverty remains the common experience of billions. One half of the world’s people live on less than $2 per day. One fifth live on less than $1.1 If modern production technologies are essentially free for the taking, then why is it that so many people are still poor? The literature that we survey here contains the beginnings of an answer to this question. First, it is true that technology is the primary determinant of a country’s income. However, the most productive techniques will not always be adopted: There are selfreinforcing mechanisms, or “traps”, that act as barriers to adoption. Traps arise both from market failure and also from “institution failure”; that is, from traps within the set of institutions that govern economic interaction. Institutions – in which we include the state, legal systems, social norms, conventions and so on – are determined endogenously within the system, and may be the direct cause of poverty traps; or they may interact with market failure, leading to the perpetuation of an inefficient status quo. There is no consensus on the view that we put forward. Some economists argue that the primary suspect for the unfortunate growth record of the least developed countries should be bad domestic policy. Sound governance and free market forces are held to be not only necessary but also sufficient to revive the poor economies, and to catalyze their convergence. Because good policy is available to all, there are no poverty traps. The idea that good policy and the invisible hand are sufficient for growth is at least vacuously true, in the sense that an all-seeing and benevolent social planner who completes the set of markets can succeed where developing country governments have failed. But this is not a theory of development, and of course benevolent social planners are not what the proponents of good governance and liberalization have in mind. Rather, their argument is that development can be achieved by the poor countries if only governments allow the market mechanism to function effectively – to get the prices right – and permit economic agents to fully exploit the available gains from trade. This re-
1 Figures are based on Chen and Ravallion (2001). Using national surveys they calculate a total head-count for the $1 and $2 poverty lines of 1.175 and 2.811 billion respectively in 1998. Their units are 1993 purchasing power adjusted US dollars.
298
C. Azariadis and J. Stachurski
quires not just openness and non-distortionary public finance, but also the enforcement of property rights and the restraint of predation.2 In essence, this is the same story that the competitive neoclassical benchmark economy tells us: Markets are complete, entry and exit is free, transaction costs are negligible, and technology is convex at an efficient scale relative to the size of the market. As a result, the private and social returns to production and investment are equal. A complete set of “virtual prices” ensures that all projects with positive net social benefit are undertaken. Diminishing returns to the set of reproducible factor inputs implies that when capital is scarce the returns to investment will be high. The dynamic implications of this benchmark were summarized by Solow (1956), Cass (1965), and Koopmans (1965). Even for countries with different endowments, the main conclusion is convergence. There are good reasons to expect this benchmark will have relevance in practice. The profit motive is a powerful force. Inefficient practices and incorrect beliefs will be punished by lost income. Further, at least one impetus shaping the institutional environment in which the market functions is the desire to mitigate or correct perceived social problems; and one of the most fundamental of all social problems is scarcity. Over time institutions have often adapted so as to relieve scarcity by addressing sources of market inefficiency.3 In any case, the intuition gained from studying the neoclassical model has been highly influential in the formulation of development policy. A good example is the structural adjustment programs implemented by the International Monetary Fund. The key components of the Enhanced Structural Adjustment Facility – the centerpiece of the IMF’s strategy to aid poor countries and promote long run growth from 1987 to 1999 – were prudent macroeconomic policies and the liberalization of markets. Growth, it was hoped, would follow automatically. Yet the evidence on whether or not non-distortionary policies and diminishing returns to capital will soon carry the poor to opulence is mixed. Even relatively well governed countries have experienced little or no growth. For example, Mali rates as “free” in recent rankings by Freedom House. Although not untroubled by corruption, it scores well in measures of governance relative to real resources [Radlet (2004), Sachs et al. (2004)]. Yet Mali is still desperately poor. According to a 2001 UNDP report, 70% of the population lives on less than $1 per day. The infant mortality rate is 230 per 1000 births, and household final consumption expenditure is down 5% from 1980. Mali is not an isolated case. In fact for all of Africa Sachs et al. (2004) argue that With highly visible examples of profoundly poor governance, for example in Zimbabwe, and widespread war and violence, as in Angola, Democratic Republic of Congo, Liberia, Sierra Leone and Sudan, the impression of a continent-wide governance crisis is understandable. Yet it is wrong. Many parts of Africa are well 2 Development theory then reduces to Adam Smith’s famous and compelling dictum, that “Little else is
requisite to carry a state to the highest degree of opulence from the lowest barbarism but peace, easy taxes, and a tolerable administration of justice”. 3 See Greif, Milgrom and Weingast (1994) for one of many possible examples.
Ch. 5: Poverty Traps
299
governed, and yet remain mired in poverty. Governance is a problem, but Africa’s development challenges are much deeper. There is a further, more subtle, problem with the “no poverty traps” argument. While the sufficiency of good policy and good governance for growth is still being debated, what can be said with certainty is that they are both elusive. The institutions that determine governance and other aspects of market interaction are difficult to reform. Almost everyone agrees that corruption is bad for growth, and yet corruption remains pervasive. Some institutions important to traditional societies have lingered, inhibiting the transition to new techniques of production. The resistance of norms and institutions to change is one reason why the outcome of liberalization and governance focused adjustment lending by the IMF has often been disappointing. To put the problem more succinctly, the institutional framework in which market interaction takes place is not implemented “from above” [Hoff (2000)]. Rather it is determined within the system. This includes the formal, legalistic aspects of the framework, but is particularly true for the informal aspects, such as social norms and conventions. The above considerations lead us back to poverty traps. First, numerous deviations from the neoclassical benchmark generate market failure. Because of these failures, good technologies are not always adopted, and productive investments are not always undertaken. Inefficient equilibria exist. Second, institutions are not always simple choice variables for benevolent national planners. Bounded rationality, imperfect information, and costly transactions make institutions and other “rules of the game” critical to economic performance; and the equilibria for institutions may be inefficient. Moreover, these inefficient equilibria have a bad habit of reinforcing themselves. Corrupt institutions can generate incentives which reward more corruption. Workers with imperfectly observed skills in an unskilled population may be treated as low skilled by firms, and hence have little incentive to invest large sums in education. Low demand discourages investment in increasing returns technology, which reduces productivity and reinforces low demand. That these inefficient outcomes are self-reinforcing is important – were they not then presumably agents would soon make their way to a better equilibrium. Potential departures from the competitive neoclassical benchmark which cause market failure are easy to imagine. One is increasing returns to scale, both internal and external. Increasing returns matter because development is almost synonymous with industrialization, and with the adoption of modern production techniques in agriculture, manufacturing and services. These modern techniques involve both fixed costs – internal economies – and greater specialization of the production process, the latter to facilitate application of machines. The presence of fixed costs for a given technology is more troubling for the neoclassical benchmark in poor countries because there market scale is relatively small. If markets are small, then the neoclassical assumption that technologies are convex at an efficient scale may be violated. The same point is true for market scale and specializa-
300
C. Azariadis and J. Stachurski
tion, in the sense that for poor countries a given increase in market scale may lead to considerably more opportunity to employ indirect production.4 Another source of increasing returns follows from the fact that modern production techniques are knowledge-intensive. As Romer (1990) has emphasized, the creation of knowledge is associated with increasing returns for several reasons. First, knowledge is non-rival and only partially excludable. Romer’s key insight is that in the presence of productive non-rival inputs, the entire replication-based logical argument for constant returns to scale immediately breaks down. Thus, knowledge creation leads to positive technical externalities and increasing returns. Second, new knowledge tends in the aggregate to complement existing knowledge. If scale economies, positive spillovers and other forms of increasing returns are important, then long run outcomes may not coincide with the predictions of the neoclassical benchmark. The essence of the problem is that when returns are increasing a rise in output lowers unit cost, either for the firm itself or for other firms in the industry. This sets in motion a chain of positive self-reinforcement. Lower unit cost encourages production, which further lowers unit cost, and so on. Such positive feedbacks can strongly reinforce either poverty or development. Another deviation from the competitive neoclassical benchmark that we discuss at length is failure in credit and insurance markets. Markets for loans and insurance suffer more acutely than most from imperfections associated with a lack of complete and symmetric information, and with all the problems inherent in anonymous trading over time. Borrowers may default or try not to pay back loans. The insured may become lax in protecting their own possessions. One result of these difficulties is that lenders usually require collateral from their borrowers. Collateral is one thing that the poor always lack. As a result, the poor are credit constrained. This can lead to an inefficient outcome which is self-reinforcing: Collateral is needed to borrow funds. Funds are needed to take advantage of economic opportunities – particularly those involving fixed costs. The ability to take advantage of opportunities determines income; and through income is determined the individual’s wealth, and hence their ability to provide collateral. Thus the poor lack access to credit markets, which is in turn the cause of their own poverty. An important aspect of this story for us is that many modern sector occupations and production techniques have indivisibilities which are not present in subsistence farming, handicraft production or other traditional sector activities. Examples include projects
4 Domestic markets are small in many developing countries, despite the possibility of international trade. In tropical countries, for example, roads are difficult to build and expensive to maintain. In Sub-Saharan Africa, overland trade with European and other markets is cut off by the Sahara. At the same time, most Sub-Saharan Africans live in the continent’s interior highlands, rather than near the coast. To compound matters, very few rivers from the interior of this part of the continent are ocean-navigable, in contrast to the geography of North America, say, or Europe [Limao and Venables (2001), Sachs et al. (2004)]. The potential for international trade to mitigate small market size is thus far lower than for a country with easy ocean access, such as Singapore or the UK.
Ch. 5: Poverty Traps
301
requiring fixed costs, or those needing large investments in human capital such as education and training. The common thread is that through credit constraints the uptake of new technologies is inhibited. With regards to insurance, it has been noted that – combined with limited access to credit – a lack of insurance is more problematic for the poor than the rich, because the poor cannot self-insure by using their own wealth. As a result, a poor person wishing to have a smooth consumption path may be forced to choose activities with low variance in returns, possibly at the cost of lower mean. Over time, lower mean income leads to more poverty. Credit and insurance markets are not the only area of the economy where limited information matters. Nor is lack of information the only constraint on economic interaction: The world we seek to explain is populated with economic actors who are boundedly rational, not rational. The fact that people are neither all-knowing nor have unlimited mental capability is important to us for several other reasons. One is that transactions become costly; and this problem is exacerbated as societies become larger and transactions more impersonal. Interaction with large societies requires more information about more people, which in turn requires more calculation and processing [North (1993, 1995)]. Second, if we concede that agents are boundedly rational then we must distinguish between the objective world and each agent’s subjective interpretation of the world. These interpretations are formed on the basis of individual and local experience, of individual inference and deduction, and of the intergenerational transmission of knowledge, values and customs. The product of these inputs is a mental model or belief system which drives, shapes and governs individual action [Simon (1986), North (1993)]. These two implications of bounded rationality are important. The first (costly transactions) because when transactions are costly institutions matter. The second (local mental models and subjective beliefs) because these features of different countries and economies shape their institutions. In this survey we emphasize two related aspects of institutions and their connection to poverty traps. The first is that institutions determine how well inefficiencies arising within the market are resolved. A typical example would be the efforts of economic and political institutions to solve coordination failure in a given activity resulting from some form of complementary externalities. The second is that institutions themselves can have inefficient equilibria. Moreover, institutions are path dependent. In the words of Paul A. David, they are the “carriers of history” [David (1994)]. Why are institutions characterized by multiple equilibria and path dependence? Although human history often shows a pattern of negotiation towards efficient institutions which mitigate the cost of transactions and overcome market failure, it is also true that institutions are created and perpetuated by those with political power. As North (1993, p. 3) has emphasized, “institutions are not necessarily or even usually created to be socially efficient; rather they, or at least the formal rules, are created to serve the interests of those with the bargaining power to create new rules”.
302
C. Azariadis and J. Stachurski
Moreover, the institutional framework is path dependent because those who currently hold power almost always have a stake in its perpetuation. Consider for example the current situation in Burundi, which has been mired in civil war since its first democratically elected president was assassinated in 1993. The economic consequences have not been efficient. Market-based economic activity has collapsed along with income. Life expectancy has fallen from 54 years in 1992 to 41 in 2000. Household final consumption expenditure is down 35% from 1980. Nevertheless, the military elite have much to gain from continuation of the war. The law of the gun benefits those with most guns. Curfew and identity checks provide opportunities for extortion. Military leaders continue to subvert a peace process that would lead to reform of the army. Path dependence is strengthened by positive feedback mechanisms which reinforce existing institutions. For example, the importance of strong property rights for growth has been extensively documented. Yet Acemoglu, Johnson and Robinson (2005, this volume) document how in Europe during the Middle Ages monarchs consistently failed to ensure property rights for the general population. Instead they used arbitrary expropriation to increase their wealth and the wealth of their allies. Increased wealth closed the circle of causation by reinforcing their own power. Engerman and Sokoloff (2005) discuss how initial inequality in some of Europe’s colonial possessions led to policies which hindered broad participation in market opportunities and strengthened the position of a small elite. Such policies tended to reinforce existing inequality (while acting as a break on economic growth). Path dependence is also inherent in the way that informal norms form the foundations of community adherence to legal stipulations. While the legal framework can be changed almost instantaneously, social norms, conventions and other informal institutions are invariably persistent (otherwise they could hardly be conventions). Often legislation is just the first step a ruling body must take when seeking to alter the de facto rules of the game.5 Finally, bounded rationality can be a source of self-reinforcing inefficient outcomes independent of institutions. For example, even in an otherwise perfect market a lack of global knowledge can cause agents to choose an inefficient technology, which is then reinforced by herd effects.6 When there are market frictions or nonconvexities such outcomes may be exacerbated. For example, if technology is nonconvex then initial poor choices by boundedly rational agents can be locked in Arthur (1994). In summary, the set of all self-reinforcing mechanisms which can potentially cause poverty is large. Even worse, the different mechanisms can interact, and reinforce one another. Increasing returns may cause investment complementarities and hence coordination failure, which is then perpetuated by pessimistic beliefs and conservative
5 For example, Transparency International’s 2004 Global Corruption Report notes that in Zambia courts
have been reluctant to hand down custodial sentences to those convicted of corruption, “principally because it was felt that white-collar criminals did not deserve to go to jail”. (Emphasis added.) 6 This example is due to Karla Hoff.
Ch. 5: Poverty Traps
303
institutions. Rent-seeking and corruption may discourage investment in new technology, which lowers expected wages for skilled workers, decreasing education effort and hence the pool of skilled workers needed by firms investing in technology. The disaffected workers may turn to rent-seeking. Positive feedbacks reinforce other feedbacks. In these kinds of environments the relevance of the neoclassical benchmark seems tenuous at best. Our survey of poverty traps proceeds as follows. Section 2 reviews key development facts. Section 3 considers several basic models associated with persistent poverty, and their implications for dynamics and the data. Section 4 looks at the empirics of poverty traps. Our survey of microfoundations is in Sections 5–8. Section 9 concludes. There are already a number of surveys on poverty traps, including two by the first author [Azariadis (1996, 2005)]. The surveys by Hoff (2000) and Matsuyama (1995, 1997) are excellent, as is Easterly (2001). See in addition the edited volumes by Bowles, Durlauf and Hoff (2005) and Mookherjee and Ray (2001). Parente and Prescott (2005) also focus on barriers to technology adoption as an explanation of cross-country variation in income levels. In their analysis institutions are treated as exogenous. 2. Development facts In Section 2.1 we briefly review key development facts, focusing on the vast and rising differences in per capita income across nations. Section 2.2 reminds the reader how these disparities came about by quickly surveying the economic history behind income divergence. 2.1. Poverty and riches What does it mean to live on one or two dollars per day? Poverty translates into hunger, lack of shelter, illness without medical attention. Calorie intake in the poorest countries is far lower than in the rich. The malnourished are less productive and more susceptible to disease than those who are well fed. Infant mortality rates in the poorest countries are up to 40 or 50 times higher than the OECD average. Many of the common causes, such as pneumonia or dehydration from diarrhea, cost very little to treat. The poor are more vulnerable to events they cannot control. They are less able to diversify their income sources. They are more likely to suffer from famine, violence and natural disasters. They have lower access to credit markets and insurance, with which to smooth out their consumption. Their children risk exploitation, and are less likely to become educated. The plight of the poor is even more striking when compared to the remarkable wealth of the rich. Measured in 1996 US dollars and adjusted for purchasing power parity, average yearly income per capita in Luxembourg for 2000 was over $46,000.7 In Tanzania, 7 Unless otherwise stated, all income data in the remainder of this section is from the Penn World Tables Version 6.1 [Heston, Summers and Aten (2002)]. Units are PPP and terms of trade adjusted 1996 US dollars.
304
C. Azariadis and J. Stachurski
by contrast, average income for 2000 was about $500. In other words, people in Luxembourg are nearly 100 times richer on average than those living in the very poorest countries.8 Luxembourg is rather exceptional in terms of per capita income, but even in the US average income is now about 70 times higher than it is in Tanzania. How has the gap between the richest and the poorest evolved over time? The answer is simple: It has increased dramatically, even in the postwar era. In 1960, per capita income in Tanzania was $478. After rising somewhat during the 1960s and 1970s it collapsed again in the 1980s. By 2000 it was $457. Many other poor countries have had similar experiences, with income hovering around the $500–1,000 mark. Meanwhile, the rich countries continued exponential growth. Income in the US grew from $12,598 in 1960 (26 times that of Tanzania) to $33,523 in 2000 (73 times). Other rich industrialized countries had similar experiences. In Australia over the same period per capita GDP rose from $10,594 to $25,641. In France it rose from $7,998 to $22,253, and in Canada from $10,168 to $26,983. Figure 1 shows how the rich have gotten richer relative to the poor. The left hand panel compares an average of real GDP per capita for the 5 richest countries in the Penn World Tables with an average of the same for the 5 poorest. The comparison is at each point in time from 1960 to the year 2000. The right panel does the same comparison with groups of 10 countries (10 richest vs. 10 poorest) instead of 5. Both panels show that by these measures income disparity has widened dramatically in the postwar era, and the rate of divergence is, if anything, increasing. The vast and growing disparity in output per person shown in Figure 1 is what growth and development theorists are obliged to explain.9 2.2. A brief history of economic development How did the massive disparities in income shown in Figure 1 arise? It is worth reviewing the broad history of economic development in order to remind ourselves of key facts.10 Although the beginnings of agriculture some ten thousand years ago marked the start of rapid human progress, for most of the subsequent millennia all but a tiny fraction of humanity was poor as we now define it, suffering regularly from hunger and highly vulnerable to adverse shocks. Early improvements in economic welfare came with the rise of premodern city-states. Collective organization of irrigation, trade, communications
8 Some countries record per capita income even lower than the figure given above for Tanzania. 1997 average income in Zaire is measured at $276. Sachs et al. (2004) use the World Bank’s 2003 World Development Indicators to calculate a population-weighted average income for Sub-Saharan Africa at 267 PPP-adjusted US dollars, or 73 cents a day. 9 Of course the figure says nothing about mobility. The poor this year could be the rich next year. See Section 4.1 for some discussion of mobility. 10 The literature on origins of modern growth is too extensive to list here. See for example the monographs of Rostow (1975) and Mokyr (2002).
Ch. 5: Poverty Traps
305
Figure 1.
and security proved more conducive to production than did autarky. Handicraft manufacture became more specialized over time, and agriculture more commercial. (Already the role of increasing returns and the importance of institutions are visible here.) While such city-states and eventually large empires rose and fell over time, and the wealth of their citizens with them, until the last few hundred years no state successfully managed the transition to what we now call modern, self-sustaining growth. Increased wealth was followed by a rise in population. Malthusian pressure led to famine and disease. The overriding reason for lack of sustained growth was that in the premodern world production technology improved only slowly. While the scientific achievements of the ancient Mediterranean civilizations and China were remarkable, in general there was little attempt to apply science to the economic problems of the peasants. Scientists and practical people had only limited interaction. Men and women of ability usual found
306
C. Azariadis and J. Stachurski
that service to the state – or predation against other states – was more rewarding than entrepreneurship and invention. Early signs of modern growth appeared in Western Europe around the middle of the last millennium. Science from the ancient world had been preserved, and now began to be extended. The revolutionary ideas of Copernicus led to intensive study of the natural world and its regularities. The printing press and movable type dramatically changed the way ideas were communicated. Innovations in navigation opened trade routes and new lands. Gunpowder and the cannon swept away local fiefdoms based on feudal castles. These technological innovations led to changes in institutions. The weakening of local fiefdoms was followed in many countries by a consolidation of central authority, which increased the scale of markets and the scope for specialization.11 Growing trade with the East and across the Atlantic produced a rich and powerful merchant class, who subsequently leveraged their political muscle to gain strengthened property and commercial rights. Increases in market size, institutional reforms and progress in technology at first lead to steady but unspectacular growth in incomes. In 1820 the richest countries in Europe had average per capita incomes of around $1,000 to $1,500 – some two or three times that of the poorest countries today. However, in the early 19th Century the vast majority of people were still poor. In this survey we compare productivity in the poor countries with the economic triumphs of the rich. Richness in our sense begins with the Industrial Revolution in Britain (although the rise in incomes was not immediate) and, subsequently, the rest of Western Europe. Industrialization – the systematic application of modern science to industrial technology and the rise of the factory system – led to productivity gains entirely different in scale from those in the premodern world. In terms of proximate causes, the Industrial Revolution in Britain was driven by a remarkable revolution in science that occurred during the period from Copernicus through to Newton, and by what Mokyr (2002) has called the “Industrial Enlightenment”, in which traditional artisanal practices were systematically surveyed, cataloged, analyzed and generalized by application of modern science. Critical to this process was the interactions of scientists with each other and with the inventors and practical men who sought to profit from innovation. Science and invention led to breakthroughs in almost all areas of production; particularly transportation, communication and manufacturing. The structure of the British economy was massively transformed in a way that had never occurred before. Employment in agriculture fell from nearly 40% in 1820 to about 12% in 1913 (and to 2.2% in 1992). The stock of machinery, equipment and non-residential structures per worker increased by a factor of five between 1820 and 1890, and then doubled again by 1913.
11 For example, in 1664 Louis XIV of France drastically reduced local tolls and unified import customs. In
1707 England incorporated Scotland into its national market. Russia abolished internal duties in 1753, and the German states instituted similar reforms in 1808.
Ch. 5: Poverty Traps
307
The literacy rate also climbed rapidly. Average years of education increased from 2 in 1820 to 4.4 in 1870 and 8.8 in 1913 [Maddison (1995)]. As a result of these changes, per capita income in the UK jumped from about $1,700 in 1820 to $3,300 in 1870 and $5,000 in 1913. Other Western European countries followed suit. In the Netherlands, income per capita grew from $1,600 in 1820 to $4,000 in 1913, while for Germany the corresponding figures are $1,100 and $3,900.12 Looking forward from the start of the last century, it might have seemed likely that these riches would soon spread around the world. The innovations and inventions behind Britain’s productivity miracle were to a large extent public knowledge. Clearly they were profitable. Adaptation to new environments is not costless, but nevertheless one suspects it was easy to feel that already the hard part had been done. Such a forecast would have been far too optimistic. Relatively few countries besides Western Europe and its off-shoots have made the transition to modern growth. Much of the world remains mired in poverty. Among the worst performers are Sub-Saharan Africa and South Asia, which together account for some 70% of the 1.2 billion people living on less than $1 per day. But poverty rates are also high in East Asia, Latin America and the Carribean. Why is it that so many countries are still poorer than 19th Century Britain? Surely the different outcomes in Britain and a country such as Mali can – at least from a modeler’s perspective – be Pareto ranked. What deviation from the neoclassical benchmark is it that causes technology growth in these countries to be retarded, and poverty to persist?
3. Models and definitions We begin our attempt to answer the question posed at the end of the last section with a review of the convex neoclassical growth model. It is appropriate to start with this model because it is the benchmark from which various deviations will be considered. Section 3.2 explains why the neoclassical model cannot explain the vast differences in income per capita between the rich and poor countries. Section 3.3 introduces the first of two “canonical” poverty trap models. These models allow us to address issues common to all such models, including dynamics and implications for the data. Section 3.4 introduces the second. 3.1. Neoclassical growth with diminishing returns The convex neoclassical model [Solow (1956)] begins with an aggregate production function of the form Yt = Ktα (At Lt )1−α ξt+1 ,
α ∈ (0, 1),
12 The figures are from Maddison (1995). His units are 1990 international dollars.
(1)
308
C. Azariadis and J. Stachurski
where Y is output of a single composite good, A is a productivity parameter, K is the aggregate stock of tangible and intangible capital, L is a measure of labor input, and ξ is a shock. In this formulation the sequence (At )t0 captures the persistent component of productivity, and (ξt )t0 is a serially uncorrelated innovation. The production function on the right-hand side of (1) represents maximum output for a given set of inputs. That output is maximal follows from competitive markets, profit seeking and free entry. (Implicit is the assumption of no significant indivisibilities or nonconvexities.) The Cobb–Douglas formulation is suggested by relative constancy of factor shares with respect to the level of worker output. Savings of tangible and intangible capital from current output occurs at constant rate s; in which case K evolves according to the rule Kt+1 = sYt + (1 − δ)Kt .
(2)
Here δ ∈ (0, 1] is a constant depreciation rate. The savings rate can be made endogenous by specifying intertemporal preferences. However the discussion in this section is purely qualitative; endogenizing savings changes little.13 If, for example, labor L is undifferentiated and grows at exogenous rate n, and if productivity A is also exogenous and grows at rate γ , then the law of motion for capital per effective worker kt := Kt /(At Lt ) is given by sktα ξt+1 + (1 − δ)kt =: G(kt , ξt+1 ), (3) θ where θ := 1 + n + γ . The evolution of output per effective worker Yt /(At Lt ) and output per capita Yt /Lt are easily recovered from (1) and (3). Because of diminishing returns, capital poor countries will extract greater marginal returns from each unit of capital stock invested than will countries with plenty of capital. The result is convergence to a long-run outcome which depends only on fundamental primitives (as opposed to beliefs, say, or historical conditions). Figure 2 shows the usual deterministic global convergence result for this model when the shock ξ is suppressed. The steady state level of capital per effective worker is kb . Figure 3 illustrates stochastic convergence with three simulated series from the law of motion (3), one with low initial income, one with medium initial income and one with high initial income. Part (a) of the figure gives the logarithm of output per effective worker, while (b) is the logarithm of output per worker. All three economies converge to the balanced growth path.14 kt+1 =
13 See, for example, Brock and Mirman (1972) or Nishimura and Stachurski (2004) for discussion of dynam-
ics when savings is chosen optimally. 14 In the simulation the sequence of shocks (ξ ) t t0 is lognormal, independent and identically distributed.
The parameters are α = 0.3, A0 = 100, γ = 0.025, n = 0, s = 0.2, δ = 0.1, and ln ξ ∼ N (0, 0.1). Here and in all of what follows X ∼ N (µ, σ ) means that X is normally distributed with mean µ and standard deviation σ .
Ch. 5: Poverty Traps
309
Figure 2.
Average convergence of the sample paths for (kt )t0 and income is mirrored by convergence in probabilistic laws. Consider for example the sequence of marginal distributions (ψt )t0 corresponding to the sequence of random variables (kt )t0 . Suppose for simplicity that the sequence of shocks is independent, identically distributed and lognormal; and that k0 > 0. It can then be shown that (a) the distribution ψt is a density for all t 1, and (b) the sequence (ψt )t0 obeys the recursion ∞ Γ k, k ψt (k) dk, for all t 1, ψt+1 k = (4) 0
where the stochastic kernel Γ in (4) has the interpretation that Γ (k, ·) is the probability density for kt+1 = G(kt , ξt+1 ) when kt is taken as given and equal to k.15 The interpretation of (4) is straightforward. It says (heuristically) that ψt+1 (k ), the probability 15 See Appendix A for details.
310
C. Azariadis and J. Stachurski
Figure 3.
that the state takes value k next period, is equal to the probability of taking value k next period given that the current state is k (that is, Γ (k, k )), summed across all k, and weighted by the probability that the current state actually takes the value k (i.e., ψt (k) dk). Here the conditional distribution Γ (k, ·) of kt+1 given kt = k is easily calculated from (3) and the familiar change-of-variable rule that if ξ is a random variable with density ϕ and Y = h(ξ ), where h is smooth and strictly monotone, then Y has density ϕ(h−1 (y)) · [dh−1 (y)/dy]. Applying this rule to (3) we get θ k − (1 − δ)k θ , Γ k, k := ϕ (5) sk α sk α where ϕ is the lognormal density of the productivity shock ξ .16
16 Precisely, z → ϕ(z) is this density when z > 0 and is equal to zero when z 0.
Ch. 5: Poverty Traps
311
All Markov processes have the property that the sequences of marginal distributions they generate satisfies a recursion in the form of (4) for some stochastic kernel Γ .17 Although the state variables usually do not themselves become stationary (due to the ongoing presence of noise), the sequence of probabilities (ψt )t0 may. In particular, the following behavior is sometimes observed: D EFINITION 3.1 (Ergodicity). Let a growth model be defined by some stochastic kernel Γ , and let (ψt )t0 be the corresponding sequence of marginal distributions generated by (4). The model is called ergodic if there is a unique probability distribution ψ ∗ supported on (0, ∞) with the property that (i) ∞ ∗ ψ k = Γ k, k ψ ∗ (k) dk for all k ; 0
and (ii) the sequence (ψt )t0 of marginal distributions for the state variable satisfies ψt → ψ ∗ as t → ∞ for all non-zero initial states.18 It is easy to see that (i) and (4) together imply that if ψt = ψ ∗ (that is, kt ∼ ψ ∗ ), then ψt+1 = ψ ∗ (that is, kt+1 ∼ ψ ∗ ) also holds (and if this is the case then kt+2 ∼ ψ ∗ follows, and so on). A distribution with this property is called a stationary distribution, or ergodic distribution, for the Markov chain. Property (ii) says that, conditional on a strictly positive initial stock of capital, the marginal distribution of the stock converges in the long run to the ergodic distribution. Under the current assumptions it is relatively straightforward to prove that the Solow process (3) is ergodic. (See the technical appendix for more details.) Figures 4 and 5 show convergence in the neoclassical model (3) to the ergodic distribution ψ ∗ . In each of the two figures an initial distribution ψ0 has been chosen arbitrarily. Since the process is ergodic, in both figures the sequence of marginal distributions (ψt )t0 converges to the same ergodic distribution ψ ∗ . This distribution ψ ∗ is determined purely by fundamentals, such as the propensity to save, the rate of capital depreciation and fertility.19 Notice in Figures 4 and 5 how initial differences are moderated under the convex neoclassical transition rule. We will see that, without convexity, initial differences often persist, and may well be amplified as the system evolves through time.
17 See Appendix A for definitions. Note that we are working here with processes that generate sequences of
densities. If the marginal distributions are not densities, and the conditional distribution contained in Γ is not a density, then the formula (4) needs to be modified accordingly. See the technical appendix. Other references include Stokey, Lucas and Prescott (1989), Futia (1982) and Stachurski (2004). 18 Convergence refers here to that of measures in the total variation norm, which in this case is just the L 1 norm. Convergence in the norm topology implies convergence in distribution in the usual sense. 19 The algorithms and code for computing marginal and ergodic distributions are available from the authors. All ergodic distributions are calculated using Glynn and Henderson’s (2001) look-ahead estimator. Marginals are calculated using a variation of this estimator constructed by the authors. The parameters in (3) are chosen – rather arbitrarily – as α = 0.3, γ = 0.02, n = 0, s = 0.2, δ = 1, and ln ξ ∼ N (3.6, 0.11).
312
C. Azariadis and J. Stachurski
Figure 4.
3.2. Convex neoclassical growth and the data The convex neoclassical growth model described in the previous section predicts that per capita incomes will differ across countries with different rates of physical and human capital formation or fertility. Can the model provide a reasonable explanation then for the fact that per capita income in the US is more than 70 times that in Tanzania or Malawi? The short answer to this question is no. First, rates at which people accumulate reproducible factors of production or have children (fertility rates) are endogenous – in fact they are choice variables. To the extent that factor accumulation and fertility are important, we need to know why some individuals and societies make choices that lead them into poverty. For poverty is suffering, and, all things being equal, few people will choose it.
Ch. 5: Poverty Traps
313
Figure 5.
This same observation leads us to suspect that the choices facing individuals in rich countries and those facing individuals in poor countries are very different. In poor countries, the choices that collectively would drive modern growth – innovation, investment in human and physical capital, etc. – must be perceived by individuals as worse than those which collectively lead to the status quo.20 A second problem for the convex neoclassical growth model as an explanation of level differences is that even when we regard accumulation and fertility rates as exogenous, they must still account for all variation in income per capita across countries. However, as many economists have pointed out, the differences in savings and fertility rates are not large enough to explain real income per capita ratios in the neighborhood 20 For this reason, endogenizing savings by specifying preferences is not very helpful, because to get poverty
in optimal growth models we must assume that the poor are poor because they prefer poverty.
314
C. Azariadis and J. Stachurski
of 70 or 100. A model ascribing output variation to these few attributes alone is insufficient. A cotton farmer in the US does not produce more cotton than a cotton farmer in Mali simply because he has saved more cotton seed. The production techniques used in these two countries are utterly different, from land clearing to furrowing to planting to irrigation and to harvest. A model which does not address the vast differences in production technology across countries cannot explain the observed differences in output. Let us very briefly review the quantitative version of this argument.21 To begin, recall the aggregate production function (1), which is repeated here for convenience: Yt = Ktα (At Lt )1−α ξt+1 .
(6) shock.22
All of the components are more or less observable besides At and the Hall and Jones (1999) conducted a simple growth accounting study by collecting data on the observable components for the year 1988. They calculate that the geometric average of output per worker for the 5 richest countries in their sample was 31.7 times that of the 5 poorest countries. Taking L to be a measure of human capital, variation in the two inputs L and K contributed only factors of 2.2 and 1.8 respectively. This leaves all the remaining variation in the productivity term A.23 This is not a promising start for the neoclassical model as a theory of level differences. Essentially, it says that there is no single map from total inputs to aggregate output that holds for every country. Why might this be the case? We know that the aggregate production function is based on a great deal of theory. Output is maximal for a given set of inputs because of perfect competition among firms. Free entry, convex technology relative to market size, price taking and profit maximization mean that the best technologies are used – and used efficiently. Clearly some aspect of this theory must deviate significantly from reality. Now consider how this translates into predictions about level differences in income per capita. When the shock is suppressed (ξt = 1 for all t), output per capita converges to the balanced path yt :=
Yt = At (s/κ)α/(1−α) , Lt
(7)
where κ := n + γ + δ.24 Suppose at first that the path for the productivity residual is the j same in all countries. That is, Ait = At for all i, j and t. In this case, the ratio of output 21 The review is brief because there are many good sources. See, for example, Lucas (1990), King and Rebelo
(1993), Prescott (1998), Hall and Jones (1999) or Easterly and Levine (2000). 22 The parameter α is the share of capital in the national accounts. Human capital can be estimated by collect-
ing data on total labor input, schooling, and returns to each year of schooling as a measure of its productivity. 23 The domestic production shocks (ξ ) t t0 are not the source of the variation. This is because they are very
small relative to the differences in incomes across countries, and, by definition, not persistent. (Recall that in our model they are innovations to the permanent component (At )t0 .) 24 When considering income levels it is necessary to assume that countries are in the neighborhood of the balanced path, for this is where the model predicts they will be. Permitting them to be “somewhere else” is not a theory of income level variation.
Ch. 5: Poverty Traps
315
Figure 6.
per capita in country i relative to that in country j is constant and equal to yi = yj
si κ j sj κ i
α/(1−α) .
(8)
The problem for the neoclassical model is that the term inside the brackets is usually not very large. For example, if we compare the US and Tanzania, say, and if we identify capital with physical capital, then average investment as a fraction of GDP between 1960 and 2000 was about 0.2 in the US and 0.24 in Tanzania. (Although the rate in Tanzania varied a great deal around this average. See Figure 6.) The average population j growth rates over this period were about 0.01 and 0.03 respectively. Since Ait = At for all t we have γ i = γ j . Suppose that this rate is 0.02, say, and that δ i = δ j = 0.05. This gives s i κ j /(s j κ i ) 1. Since payments to factors of production suggest that α/(1 − α)
316
C. Azariadis and J. Stachurski
is neither very large nor very small, output per worker in the two countries is predicted to be roughly equal. This is only an elementary calculation. The computation of investment rates in Tanzania is not very reliable. There are issues in terms of the relative ratios of consumption and investment good prices in the two countries which may distort the data. Further, we have not included intangible capital – most notably human capital. The rate of investment in human capital and training in the US is larger than it is in Tanzania. Nevertheless, it is difficult to get the term in (8) to contribute a factor of much more than 4 or 5 – certainly not 70.25 However the calculations are performed, it turns out that to explain the ratio of incomes in countries such as Tanzania and the US, productivity residuals must absorb most of the variation. In other words, the convex neoclassical growth model cannot be reconciled with the cross-country income data unless we leave most of the variation in income to an unexplained residual term about which we have no quantitative theory. And surely any scientific theory can explain any given phenomenon by adopting such a strategy. Different authors have made this same point in different ways. Lucas (1990) notes that if factor input differences are large enough to explain cross-country variations in income, the returns to investment in physical and human capital in poor countries implied by the model will be huge compared to those found in the rich. They are not. Also, productivity residuals are growing quickly in countries like the US.26 On the other hand, in countries like Tanzania, growth in the productivity residual has been very small.27 Yet the convex neoclassical model provides no theory on why these different rates of growth in productivity should hold. On balance, the importance of productivity residuals suggests that the poor countries are not rich because for one reason or another they have failed or not been able to adopt modern techniques of production. In fact production technology in the poorest countries is barely changing. In West Africa, for example, almost 100% of the increase in per capita food output since 1960 has come from expansion of harvest area [Baker (2004)]. On the other hand, the rich countries are becoming ever richer because of continued innovation.
25 See in particular Prescott (1998) for detailed calculations. He concludes that convex neoclassical growth
theory “fails as a theory of international income differences, even after the concept of capital is broadened to include human and other forms of intangible capital. It fails because differences in savings rates cannot account for the great disparity in per capita incomes unless investment in intangible capital is implausibly large.” 26 One can compute this directly, or infer it from the fact that interest rates in the US have shown no secular trend over the last century, in which case transitional dynamics can explain little, and therefore growth in output per worker and growth in the residual can be closely identified [King and Rebelo (1993)]. 27 Again, this can be computed directly, or inferred from the fact that if it had been growing at a rate similar to the US, then income in Tanzania would have been at impossibly low levels in the recent past [Pritchett (1997)].
Ch. 5: Poverty Traps
317
Of course this only pushes the question one step back. Technological change is only a proximate cause of diverging incomes. What economists need to explain is why production technology has improved so quickly in the US or Japan, say, and comparatively little in countries such as Tanzania, Mali and Senegal. We end this section with some caveats. First, the failure of the simple convex neoclassical model does not imply the existence of poverty traps. For example, we may discover successful theories that predict very low levels of the residual based on exogenous features which tend to characterize poor countries. (Although it may turn out that, depending on what one is prepared to call exogenous, the map from fundamentals to outcomes is not uniquely defined. In other words, there are multiple equilibria. In Section 4.2 some evidence is presented on this point.) Further, none of the discussion in this section seeks to deny that factor accumulation matters. Low rates of factor accumulation are certainly correlated with poor performance, and we do not wish to enter the “factor accumulation versus technology” debate – partly because this is viewed as a contest between neoclassical and “endogenous” growth models, which is tangential to our interests, and partly because technology and factor accumulation are clearly interrelated: technology drives capital formation and investment boosts productivity.28 Finally, it should be emphasized that our ability to reject the elementary convex neoclassical growth model as a theory of level differences between rich and poor countries is precisely because of its firm foundations in theory and excellent quantitative properties. All of the poverty trap models we present in this survey provide far less in terms of quantitative, testable restrictions that can be confronted with the data. The power of a model depends on its falsifiability, not its potential to account for every data set. 3.3. Poverty traps: historical self-reinforcement How then are we to explain the great variation in cross-country incomes such as shown in Figure 1? In the introduction we discussed some deviations from the neoclassical benchmark which can potentially account for this variation by endogenously reinforcing small initial differences. Before going into the specifics of different feedback mechanisms, this section formulates the first of two abstract poverty trap models. For both models a detailed investigation of microfoundations is omitted. Instead, our purpose is to establish a framework for the questions poverty traps raise about dynamics, and for their observable implications in terms of the cross-country income data. The first model – a variation on the convex neoclassical growth model discussed in Section 3.1 – is loosely based on Romer (1986) and Azariadis and Drazen (1990). It exemplifies what Mookherjee and Ray (2001) have called historical self-reinforcement, a process whereby initial conditions of the endogenous variables can shape long run
28 However, as we stressed at the beginning of this section, to the extent that factor accumulation is important
it may in fact turn out that low accumulation rates are mere symptoms of poverty, not causes.
318
C. Azariadis and J. Stachurski
outcomes. Leaving aside all serious complications for the moment, let us fix at s > 0 the savings rate, and at zero the rates of exogenous technological progress γ and population growth n. Let all labor be undifferentiated and normalize its total mass to 1, so that k represents both aggregate capital and capital per worker. Suppose that the productivity parameter A can vary with the stock of capital. In other words, A is a function of k, and aggregate returns kt → A(kt )ktα are potentially increasing.29 The law of motion for the economy is then kt+1 = sA(kt )ktα ξt+1 + (1 − δ)kt .
(9)
Depending on the specification of the relationship between k and productivity, many dynamic paths are possible. Some of them will lead to poverty traps. Figure 7 gives examples of potential dynamic structures. For now the shock ξ is suppressed. The x-axis is current capital kt and the y-axis is kt+1 . In each case the plotted curve is just the righthand side of (9), all with different maps k → A(k). In part (a) of the figure the main feature is non-ergodic dynamics: long run outcomes depend on the initial condition. Specifically, there are two local attractors, the basins of attraction for which are delineated by the unstable fixed point kb . Part (b) is also nonergodic. It shows the same low level attractor, but now no high level attractor exists. Beginning at a state above kb leads to unbounded growth. In part (c) the low level attractor is at zero. The figure in part (d) looks like an anomaly. Since the dynamics are formally ergodic, many researchers will not view this structure as a “poverty trap” model. Below we argue that this reading is too hasty: the model in (d) can certainly generate the kind of persistent-poverty aggregate income data we are hoping to explain. In order to gain a more sophisticated understanding, let us now look at the stochastic dynamics of the capital stock. Deterministic dynamics are of course a special case of stochastic dynamics (with zero-variance shocks) but as in the case of the neoclassical model above, let us suppose that (ξt )t0 is independently and identically lognormally distributed, with ln ξ ∼ N (µ, σ ) and σ > 0. It then follows that the sequence of marginal distributions (ψt )t0 for the capital stock sequence (kt )t0 again obeys the recursion (4) where the stochastic kernel Γ is now 1 k − (1 − δ)k , Γ k, k := ϕ (10) sA(k)k α sA(k)k α with ϕ the lognormal density on (0, ∞) and zero elsewhere. All of the intuition for the recursion (4) and the construction of the stochastic kernel (10) is exactly the same as the neoclassical case. 29 In Romer (1986), for example, private investment generates new knowledge, some of which enters the
public domain and can be used by other firms. In Azariadis and Drazen (1990) there are spillovers from human capital formation. See also Durlauf (1993) and Zilibotti (1995). See Matsuyama (1997) and references for discussion of how investment may feed back via pecuniary externalities into specialization and hence productivity. Our discussion of microfoundations begins in Section 5.
Ch. 5: Poverty Traps
319
Figure 7.
How do the marginal distributions of the nonconvex growth model evolve? The following result gives the answer for most cases we are interested in. P ROPOSITION 3.1. Let (ξt )t0 be an independent sequence with ln ξt ∼ N (µ, σ ) for all t, and let σ > 0. If k → A(k) satisfies the regularity condition 0 < inf A(k) sup A(k) < ∞, k
k
then the stochastic nonconvex growth model defined by (9) is ergodic.30
30 In fact we require also that k → A(k) is a Borel measurable function. But this condition is very weak
indeed. For example, k → A(k) need be neither monotone nor continuous.
320
C. Azariadis and J. Stachurski
Figure 8.
Ergodicity here refers to Definition 3.1, which, incidentally, is the standard definition used in growth theory and macroeconomics [see, for example, Brock and Mirman (1972); or Stokey, Lucas and Prescott (1989)]. In other words, there is a unique ergodic distribution ψ ∗ , and the sequence of marginal distributions (ψt )t0 converges to ψ ∗ asymptotically, independent of the initial condition (assuming of course that k0 > 0). A proof of this result is given in Appendix A. So why has a non-ergodic model become ergodic with the introduction of noise? The intuition is completely straightforward: Under our assumption of unbounded shocks there is always the potential – however small – to escape any basin of attraction. So in the long run initial conditions do not matter. (What does matter is how long this long run is, a point we return to below.) Figure 8 gives the ergodic distributions corresponding to two poverty trap models.31 Both have the same structural dynamics as the model in part (a) of Figure 7. The left 31 Regarding numerical computation see the discussion for the neoclassical case above.
Ch. 5: Poverty Traps
321
hand panels show this structure with the shock suppressed. The right hand panels show corresponding ergodic distributions under the independent lognormal shock process. Both ergodic distributions are bimodal, with modes concentrated around the deterministic local attractors. Comparing the two left hand panels, notice that although qualitatively similar, the laws of motion for Country A and Country B have different degrees of increasing returns. For Country B, the jump occurring around k = 4 is larger. As a result, the state is less likely to return to the neighborhood of the lower attractor once it makes the transition out of the poverty trap. Therefore the mode of the ergodic distribution corresponding to the higher attractor is large relative to that of Country A. Economies driven by law of motion B spend more time being rich. Convergence to the ergodic distribution in a nonconvex growth model is illustrated in Figure 9. The underlying model is (a) of Figure 7.32 As before, the ergodic distribution is bimodal. In this simulation, the initial distribution was chosen arbitrarily. Note how initial differences tend to be magnified over the medium term despite ergodicity. The initially rich diverge to the higher mode, creating the kind of “convergence club” effect already seen in ψ15 , the period 15 marginal distribution.33 It is clear, therefore, that ergodicity is not the whole story. If the support of the shock ξ is bounded then ergodicity may not hold. Moreover, even with ergodicity, historical conditions may be arbitrarily persistent. Just how long they persist depends mainly on (i) the size of the basins of attraction and (ii) the statistical properties of the shock. On the other hand, the non-zero degree of mixing across the state space that drives ergodicity is usually more realistic than deterministic models where poverty traps are absolute and can never be overcome. Indeed, we will see that ergodicity is very useful for framing empirical questions in Section 4.2. Figures 10 and 11 illustrate how historical conditions persist for individual time series generated by a model in the form of (a) of Figure 7, regardless of ergodicity. In both figures, the x-axis is time and the y-axis is (the log of) capital stock per worker. The dashed line through the middle of the figure corresponds to (the log of) kb , the point dividing the two basins of attraction in (a) of Figure 7. Both figures show the simulated time series of four economies. In each figure, all four economies are identical, apart from their initial conditions. One economy is started in the basin of attraction for the higher attractor, and three are started in that of the lower attractor.34 In the figures, the economies spend most of the time clustered in the neighborhoods of the two deterministic attractors. Economies starting in the portion of the state space 32 The specification of A(k) used in the simulation is A(k) = a exp(hΨ (k)), where a = 15, h = 0.52 and the transition function Ψ is given by Ψ (k) := (1 + exp(− ln(k/kT )/θ))−1 . The parameter kT is a “threshold”
value of k, and is set at 6.9. The parameter θ is the smoothness of the transition, and is set at 0.09. The other parameters are α = 0.3, s = 0.2, δ = 1, and ln ξ ∼ N (0, 0.1). 33 Incidentally, the change in the distributions from ψ to ψ is qualitatively quite similar to the change in 0 15 the cross-country income distribution that has been observed in the post war period. 34 The specification of A(k) is as in Figure 9, where now k = 4.1, θ = 0.2, h = 0.95, α = 0.3, s = 0.2, T δ = 1. For Figure 10 we used ln ξ ∼ N (0, 0.1), while for Figure 11 we used ln ξ ∼ N (0, 0.05).
322
C. Azariadis and J. Stachurski
Figure 9.
(the y-axis) above the threshold are attracted on average to the high level attractor, while those starting below are attracted on average to the low level attractor. For these parameters, historical conditions are important in determining outcomes over the kinds of time scales economists are interested in, even though there are no multiple equilibria, and in the limit outcomes depend only on fundamentals. In Figure 10, all three initially poor economies eventually make the transition out of the poverty trap, and converge to the neighborhood of the high attractor. Such transitions might be referred to as “growth miracles”. In these series there are no “growth disasters” (transitions from high to low). The relative likelihood of growth miracles and growth disasters obviously depends on the structure of the model – in particular, on the relative size of the basins of attraction. In Figure 10 the shock is distributed according to ln ξ ∼ N (0, 0.1), while in Figure 11 the variance is smaller: ln ξ ∼ N (0, 0.05). Notice that in Figure 11 no growth
Ch. 5: Poverty Traps
323
Figure 10.
miracles occur over this time period. The intuition is clear: With less noise, the probability of a large positive shock – large enough to move into the basin of attraction for the high attractor – is reduced, and with it the probability of escaping from the poverty trap. We now return to the model in part (d) of Figure 7, which is nonconvex, but at the same time is ergodic even in the deterministic case. This kind of structure is usually not regarded as a poverty trap model. In fact, since (d) is just a small perturbation of model (a), the existence of poverty traps is often thought to be very sensitive to parameters – a small change can cause a bifurcation of the dynamics whereby the poverty trap disappears. But, in fact, the phenomenon of persistence is more subtle. In terms of their medium run implications for cross-country income patterns, the two models (a) and (d) are very similar.
324
C. Azariadis and J. Stachurski
Figure 11.
To illustrate this, Figure 12 shows an arbitrary initial distribution and the resulting time 5 distribution for k under the law of motion given in (d) of Figure 7.35 As in all cases we have considered, the stochastic model is ergodic. Now the ergodic distribution (not shown) is unimodal, clustered around the single high level attractor of the deterministic model. Thus the long run dynamics are different to those in Figure 9. However, during the transition, statistical behavior is qualitatively the same as that for models that do have low level attractors (such as (a) of Figure 7). In ψ5 we observe amplification of initial differences, and the formation of a bimodal distribution with two “convergence clubs”. 35 The specification of A(k) is as before, where now k = 3.1, θ = 0.15, h = 0.7, α = 0.3, s = 0.2, δ = 1, T
and ln ξ ∼ N (0, 0.2).
Ch. 5: Poverty Traps
325
Figure 12.
How long is the medium run, when the transition is in progress and the distribution is bimodal? In fact one can make this transition arbitrarily long without changing the basic qualitative features of (d), such as the non-existence of a low level attractor. Its length depends on the degree of nonconvexity and the variance of the productivity shocks (ξt )t0 . Higher variance in the shocks will tend to speed up the transition. Incidentally, the last two examples have illustrated an important general principle: In economies with nonconvexities, the dynamics of key variables such as income can be highly sensitive to the statistical properties of the exogenous shocks which perturb activity in each period.36 This phenomenon is consistent with the cross-country income panel. Indeed, several studies have emphasized the major role that shocks play in de-
36 Such sensitivity is common to all dynamic systems where feedbacks can be positive. The classic example
is evolutionary selection.
326
C. Azariadis and J. Stachurski
termining the time path of economic development [cf., e.g., Easterly et al. (1993), Den Haan (1995), Acemoglu and Zilibotti (1997), Easterly and Levine (2000)].37 At the risk of some redundancy, let us end our discussion of the increasing returns model (9) by reiterating that persistence of historical conditions and formal ergodicity may easily coincide. (Recall that the time series in Figure 11 are generated by an ergodic model, and that (d) of Figure 7 is ergodic even in the deterministic case.) As a result, identifying history dependence with a lack of ergodicity can be problematic. In this survey we use a more general definition: D EFINITION 3.2 (Poverty trap). A poverty trap is any self-reinforcing mechanism which causes poverty to persist. When considering a given quantitative model and its dynamic implications, the important question to address is, how persistent are the self-reinforcing mechanisms which serve to lock in poverty over the time scales that matter when welfare is computed?38 A final point regarding this definition is that the mechanisms which reinforce poverty may occur at any scale of social and spatial aggregation, from individuals to families, communities, regions, and countries. Traps can arise not just across geographical location such as national boundaries, but also within dispersed collections of individuals affiliated by ethnicity, religious beliefs or clan. Group outcomes are then summed up progressively from the level of the individual.39 3.4. Poverty traps: inertial self-reinforcement Next we turn to our second “canonical” poverty trap model, which again is presented in a very simplistic form. (For microfoundations see Sections 5–8.) The model is static rather than dynamic, and exhibits what Mookherjee and Ray (2001) have described as inertial self-reinforcement.40 Multiple equilibria exist, and selection of a particular equilibrium can be determined purely by beliefs or subjective expectations. In the economy a unit mass of agents choose to work either in a traditional, rural sector or a modern sector. Labor is the only input to production, and each agent supplies one unit in every period. All markets are competitive. In the traditional sector returns to scale are constant, and output per worker is normalized to zero. The modern sector, however, is knowledge-intensive, and aggregate output exhibits increasing returns due perhaps to spillovers from agglomeration, or from matching and network effects. 37 This point also illustrates a problem with standard empirical growth studies. In general no information on
the shock distribution is incorporated into calculation of dynamics. 38 Mookherjee and Ray (2001) have emphasized the same point. See their discussion of “self-reinforcement
as slow convergence”. 39 This point has been emphasized by Barrett and Swallow (2003) in their discussion of “fractal” poverty
traps. 40 By “static” we mean that there are no explicitly specified interactions between separate periods.
Ch. 5: Poverty Traps
327
Figure 13.
Let the fraction of agents working in the modern sector be denoted by α. The map α → f (α) gives output per worker in the modern sector as a function of the fraction employed there. Payoffs are just wages, which equal output per worker (marginal product). Agents maximize individual payoffs taking the share α as exogenously given. We are particularly interested in the case of strategic complementarities. Here, entry into the modern sector exhibits complementarities if the payoff to entering the modern sector increases with the number of other agents already there; in other words, if f is increasing. We assume that f > 0, and also that returns in the modern sector dominate those in the traditional sector only when the number of agents in the modern sector rises above some threshold. That is, f (0) < 0 < f (1). This situation is shown in Figure 13. At the point αb returns in the two sectors are equal. Equilibrium distributions of agents are values of α such that f (α) = 0, as well as “all workers are in the traditional sector”, or “all workers are in the modern sector”
328
C. Azariadis and J. Stachurski
(ignoring adjustments on null sets). The last two of these are clearly Pareto-ranked: The equilibrium α = 0 has the interpretation of a poverty trap. Immediately the following objection arises. Although the lower equilibrium is to be called a poverty trap, is there really a self-reinforcing mechanism here which causes poverty to persist? After all, it seems that as soon as agents coordinate on the good equilibrium “poverty” will disappear. And there are plenty of occasions where societies acting collectively have put in place the institutions and preconditions for successful coordination when it is profitable to do so. Although the last statement is true, it seems that history still has a role to play in equilibrium selection. This argument has been discussed at some length in the literature, usually beginning with myopic Marshallian dynamics, under which factors of production move over time towards activities where returns are higher. In the case of our model, these dynamics are given by the arrows in Figure 13. If (α0 )t0 is the sequence of modern sector shares, and if initially α0 < αb , then αt → 0. Conversely, if α0 > αb , then αt → 1. But, as many authors have noted, this analysis only pushes the question one step back. Why should the sectoral shares only evolve slowly? And if they can adjust instantaneously, then why should they depend on the initial condition at all? What are the sources of inertia here that prevent agents from immediately coordinating on good equilibria?41 Adsera and Ray (1997) have proposed one answer. Historical conditions may be decisive if – as seems quite plausible – spillovers in the modern sector arise only with a lag. A simplified version of the argument is as follows. Suppose that the private return to working in the modern sector is rt , where now r0 = f (α0 ) and rt takes the lagged value f (αt−1 ) when t 1. Suppose also that at the end of each period agents can move costlessly between sectors. Agent j chooses location in order to maximize a discounted j sum of payoffs given subjective beliefs (αt )t0 for the time path of shares, where to be j consistent we require that α0 = α0 for all j . Clearly, if α0 < αb , then switching to or remaining in the traditional sector at the end of time zero is a dominant strategy regardless of beliefs, because r1 = f (α0 ) < f (αb ) = 0. The collective result of these individual decisions is that α1 = 0. But then α1 < αb , and the whole process repeats. Thus αt = 0 for all t 1. This outcome is interesting, because even the most optimistic set of beliefs lead to the low equilibrium when f (α0 ) < 0. To the extent that Adsera and Ray’s analysis is correct, history must always determine outcomes.42 Another way that history can re-enter the equation is if we admit some deviation from perfect rationality and perfect information. As was stressed in the introduction,
41 See, for example, Krugman (1991) or Matsuyama (1991). 42 There are a number of possible criticisms of the result, most of which are discussed in detail by the authors.
If, for example, there are congestion costs or first mover advantages, then moving immediately to the modern sector might be rational for some optimistic beliefs and specification of parameters.
Ch. 5: Poverty Traps
329
this takes us back to the role of institutions, through which history is transmitted to the present. It is reasonable to entertain such deviations here for a number of reasons. First and foremost, assumptions of complete information and perfect rationality are usually justified on the basis of experience. Rationality obtains by repeated observation, and by the punishment of deviant behavior through the carrot and stick of economic payoff. Rational expectations are justified by appealing to induction. Agents are assumed to have had many observations from a stationary environment. Laws of motion and hence conditional expectations are inferred on the basis of repeated transition sampling from every relevant state/action pair [Lucas (1986)]. When attempting to break free from a poverty trap, however, agents have most likely never observed a transition to the high level equilibrium. On the basis of what experience are they to assess its likelihood from each state and action? How will they assess the different costs or benefits? In a boundedly rational environment with limited information, outcomes will be driven by norms, institutions and conventions. It is likely that these factors are among the most important in terms of a society’s potential for successful coordination on good equilibria. In fact for some models we discuss below the equilibrium choice is not between traditional technology and the modern sector, but rather is a choice between predation (corruption) and production, or between maintaining kinship bonds and breaking them. In some sense these choices are inseparable from the social norms and institutions of the societies within which they are framed.43 The central role of institutions may not prevent rapid, successful coordination on good equilibria. After all, institutions and conventions are precisely how societies solve coordination problems. As was emphasized in the introduction, however, norms, institutions and conventions are path dependent by definition. And, in the words of Matsuyama (1995, p. 724), “coordinating expectations is much easier than coordinating changes in expectations”. Because of this, economies that start out in bad equilibria may find it difficult to break free. Why should a convention that locks an economy into a bad equilibrium develop in the first place? Perhaps this is just the role of historical accident. Or perhaps, as Sugden (1989) claims, conventions tend to spread on the basis of versatility or analogy.44 If so, the conventions that propagate themselves most successfully may be those which are most versatile or susceptible to analogy – not necessarily those which lead to “good” or efficient equilibria. Often the debate on historical conditions and coordination is cast as “history versus expectations”. We have emphasized the role of history, channeled through social norms 43 More traditional candidates for coordinating roles among the set of institutions include interventionist
states promoting industrialization through policy-based financing, or (the cultural component of) large business groups, such as Japan’s keiretsu and South Korea’s chaebol. In Section 5.2, we discuss the potential for large banks with significant market power to drive “big push” type investments by the private sector. 44 A versatile convention works reasonably well against many strategies, and hence is advantageous when facing great uncertainty. Analogy means that a rule for a particular situation is suggested by similar rules applied to different but related situations.
330
C. Azariadis and J. Stachurski
and institutions, but without intending to say that beliefs are not important. Rather, beliefs are no doubt crucial. At the same time, beliefs and expectations are shaped by history. And they in turn combine with value systems and local experience to shape norms and institutions. The latter then determine how successful different societies are in solving the particular coordination problems posed by interactions in free markets. If beliefs and expectations are shaped by history, then the “history versus expectations” dichotomy is misleading. The argument that beliefs and expectations are indeed formed by a whole variety of historical experiences has been made by many development theorists. In an experiment investigating the effects of the Indian caste system, Hoff and Pandey (2004) present evidence that individuals view the world through their own lens of “historically created social identities”, which in turn has a pronounced effect on expectations. Rostow (1990, p. 5) writes that “the value system of [traditional] societies was generally geared to what might be called a long run fatalism; that is, the assumption that the range of possibilities open to one’s grandchildren would be just about what it had been for one’s grandparents”. Ray (2003, p. 1) argues that “poverty and the failure of aspirations may be reciprocally linked in a self-sustaining trap”. Finally, experimental evidence on coordination games with multiple Pareto-ranked equilibria suggests that history is important: Outcomes are strongly path dependent. For example, Van Huyck, Cook and Battalio (1997) study people’s adaptive behavior in a generic game of this type, where multiple equilibria are generated by strategic complementarities. In each experiment, eight subjects participated in a sequence of between 15 and 40 plays. The authors find sensitivity to initial conditions, defined here as the median of the first round play. In their view, “the experiment provides some striking examples of coordination failure growing from small historical accidents”.
4. Empirics of poverty traps Casual observation of the cross-country income panel tends to suggest mechanisms which reinforce wealth or poverty. In Section 4.1 we review the main facts. Section 4.2 considers tests for the empirical relevance of poverty trap models. While the results of the tests support the hypothesis that the map from fundamentals to economic outcomes is not unique, they give no indication as to what forces might be driving multiplicity. Section 4.3 begins the difficult task of addressing this issue in a macroeconomic framework. Finally, Section 4.4 gives references to empirical tests of specific microeconomic mechanisms that can reinforce poverty at the individual or group level. 4.1. Bimodality and convergence clubs A picture of the evolving cross-country income distribution is presented in Figure 14. For both the top and bottom histograms the y-axis measures frequency. For the top pair (1960 and 1995) the x-axis is GDP per capita in 1996 PPP adjusted dollars. This is the
Ch. 5: Poverty Traps
331
Figure 14.
standard histogram of the cross-country income distribution. For the bottom pair the x-axis represents income as a fraction of the world average for that year. The single most striking feature of the absolute income histograms for 1960 and 1995 is that over this period a substantial fraction of poor countries have grown very little or not at all. At the same time, a number of middle income countries have grown rapidly, in some cases fast enough to close in on the rich. Together, these forces have caused the distribution to become somewhat thinner in the middle, with probability mass collecting at the two extremes. Such an outcome is consistent with mechanisms that accentuate differences in initial conditions, and reinforce wealth or poverty [Azariadis and Drazen (1990), Quah (1993, 1996), Durlauf and Johnson (1995), Bianchi (1997), Pritchett (1997), Desdoigts (1999), Easterly and Levine (2000)].
332
C. Azariadis and J. Stachurski
As well as observing past and present distributions, Quah (1993) also used the Penn World Tables to estimate a transition probability matrix by discretizing the state space (income per capita), treating all countries as observations from the same Markovian probability law, and measuring transition frequency. This matrix provides information on mobility. Also, by studying the ergodic distribution, and by multiplying iterations of the matrix with the current cross-country income distribution, some degree of inference can be made as to where the income distribution is heading. In his calculation, Quah uses per capita GDP relative to the world average over the period 1962 to 1984 in a sample of 118 countries. Relative income is discretized into state space S := {1, 2, 3, 4, 5} consisting of 5 “bins”, with states corresponding to values for relative GDP of 0–0.25, 0.25–0.5, 0.5–1, 1–2 and 2–∞ respectively. The transition matrix P = (pij ) is computed by setting pij equal to the fraction of times that a country, finding itself in state i, makes the transition to state j the next year. The data is assumed to be stationary, so that all of the transitions can be pooled when calculating transition probabilities. The result of this calculation [Quah (1993, p. 431)] is 0.97 0.03 0.00 0.00 0.00 0.05 0.92 0.03 0.00 0.00 P= 0.00 0.04 0.92 0.04 0.00 . 0.00 0.00 0.04 0.94 0.02 0.00 0.00 0.00 0.01 0.99 The Markov chain represented by P is easily shown to be ergodic, in the sense that there is a unique ψ ∗ ∈ P(S), the distributions on S, with the property that ψ ∗ P = ψ ∗ , and ψPt → ψ ∗ as t → ∞ for all ψ ∈ P(S).45 Quah calculates this ergodic distribution ψ ∗ to be (0.24, 0.18, 0.16, 0.16, 0.27). The ergodic distribution is quite striking, in that the world is divided almost symmetrically into two convergence clubs of rich and poor at either end of the income distribution. It is not immediately clear just how long the long run is. To get some indication, we can apply Pt to the current distribution for different values of t. Figure 15 shows the results of applying P30 to the year 2000 income distribution from the Penn World Tables. This gives a projection for the 2030 distribution. Contrasted with the 1960 distribution the prediction is strongly bimodal. As Quah himself was at pains to emphasize, the projections carried out above are only a first pass at income distribution dynamics, with many obvious problems. One is that the dynamics generated by a discretized version of a continuous state Markov chain can deviate very significantly from the true dynamics generated by the original chain, and error bounds are difficult to quantify.46 Also, since the estimation of P is purely
45 Following Markov chain convention we are treating the distributions in P(S) as row vectors. Also, Pt is
t compositions of P with itself. For more discussion of ergodicity see the technical appendix, or Stachurski (2004). 46 Compare, for example, Feyrer (2003) and Johnson (2004).
Ch. 5: Poverty Traps
333
Figure 15.
nonparametric, the projections do not contain any of the restrictions implied by growth theory. Quah (1996) addressed the first of these problems by estimating a continuous state version. In the language of this survey, he estimates a stochastic kernel Γ , of which P is a discretized representation. The estimation is nonparametric, using a Parzen-window type density smoothing technique. The kernel is suggestive of considerable persistence. Azariadis and Stachurski (2004) make some effort to address both the discretization problem and the lack of economic theory simultaneously, by estimating Γ parametrically, using a theoretical growth model. In essence, they estimate Equation (9), where k → A(k) is represented by a three-parameter logistic function. The logistic function nests a range of growth models, from the convex model in Figure 2 to the nonconvex models in Figure 7, panels (a), (b) and (d). Once the law of motion (9) is estimated, the stochastic kernel Γ is calculated via Equation (10), and the projection of distributions is computed by iterating (4).
334
C. Azariadis and J. Stachurski
Figure 16.
The resulting 2030 prediction is shown in Figure 16, with the 1960 distribution drawn above for comparison. The x-axis is log of real GDP per capita in 1996 US dollars. The 1960 density is just a smoothed density estimate using Gaussian kernels, with data from the Penn World Tables. The same data was used to estimate the parameters in the law of motion (9). As in Figure 15, a unimodal distribution gives way to a bimodal distribution. These findings do lend some support to Quah’s convergence club hypothesis. Much work remains to be done. For example, in all of the methodologies discussed above, nonstationary data is being fitted to a stationary Markov chain. This is clearly a source of bias. Furthermore, all of these models are too small, in the sense that the state space used in the predictions are only one-dimensional.47 47 In fact within each economy there are many interacting endogenous variables, only one of which is income.
Even if the process as a whole is stationary and Markov, projection of the system onto one dimension will
Ch. 5: Poverty Traps
335
4.2. Testing for existence Poverty trap models tend to be lacking in testable quantitative implications. Where there are multiple equilibria and sensitive dependence to initial conditions, outcomes are much harder to pin down than when the map from parameters to outcomes is robust and unique. This has led many economists to question the empirical significance of poverty trap models.48 In this section, we ask whether or not there is any evidence that poverty traps exist. In answering this question, one must be very careful to avoid the following circular logic: First, persistent poverty is observed. Poverty traps are then offered as the explanation. But how do we know there are poverty traps? Because (can’t you see?) poverty persists.49 This simple point needs to be kept in mind when interpreting the data with a view to assessing the empirical relevance of the models in this survey. Persistent poverty, emergent bimodality and the dispersion of cross-country income are the phenomena we seek to explain. They cannot themselves be used as proof that poverty traps explain the data. Also, a generalized convex neoclassical model can certainly be the source of bimodality and dispersion if we accept that the large differences in total factor productivity residuals across countries are due to some exogenous force, the precise nature of which is still waiting to be explained. In this competing explanation, the map from fundamentals to outcomes is unique, and shocks or historical accidents which perturb the endogenous variables can safely be ignored. The central question, then, is whether or not the poverty trap explanation of crosscountry income differentials survives if we control for the exogenous forces which determine long run economic performance. In other words, do self-reinforcing and path dependent mechanisms imply that economies populated by fundamentally similar people in fundamentally similar environments can support very different long run outcomes? What empirical support is there for such a hypothesis? One particularly interesting study which addresses this question is that of Bloom, Canning and Sevilla (2003). Their test is worth discussing in some detail. To begin, consider again the two multiple equilibria models shown in Figure 8 (Section 3.3), along with their ergodic distributions. As can be seen in the left hand panels, when the shock is suppressed both Country A and Country B have two locally stable equilibria for capital per worker – and therefore two locally stable equilibria for income. Call these two states y1∗ and y2∗ , the first of which is interpreted as the poverty trap. yield a process which is not generally Markovian. Moreover, there are interactions between countries that affect economic performance, and these interactions are important. A first-best approach would be to treat the world economy as an N × M-dimensional Markov process, where N is the number of countries, and M is the number of endogenous variables in each country. One would then estimate the stochastic kernel Γ for N ×M N ×M this process, a map from R+ × R+ → [0, ∞). Implications for the cross-country income distribution could be calculated by computing marginals. 48 See Matsuyama (1997) for more discussion of this point. 49 Recall Karl Popper’s famous tale about Neptune and the sea.
336
C. Azariadis and J. Stachurski
In general, y1∗ and y2∗ will depend on the vector of exogenous fundamentals, which determine the exact functional relationships in the model, and hence become parameters in the law of motion. Let this vector be denoted by x. Consider a snapshot of the economy at some point in time t. We can write income per capita as ∗ y1 (x) + u1 with probability p(x); y= (R2) y2∗ (x) + u2 with probability 1 − p(x). Here p(x) is the probability that the country in question is in the basin of attraction for the lower equilibrium y1∗ (x) at time t. This probability is determined by the time t marginal distribution of income. The shock ui represents deviation from the deterministic attractor at time t. Figure 8 helps to illustrate how y1∗ and y2∗ might depend on the exogenous variables. Imagine that Countries A and B have characteristics xA and xB respectively. These different characteristics account for the different shapes of the laws of motion shown in the left-hand side of the figure. As drawn, y2∗ (xA ), the high level attractor for Country A, is less than y2∗ (xB ), the high level attractor for Country B, while y1∗ (xA ) and y1∗ (xB ) are roughly equal. In addition, we can see how the probability p(x) of being in the poverty trap basin depends on these characteristics. For time t sufficiently large, ergodicity means that the time t marginal distribution – which determines this probability – can be identified with the ergodic distribution. The ergodic distribution in turn depends on the underlying structure, which depends on x. This is illustrated by the different sizes of the distribution modes for Countries A and B in Figure 8. For Country A the left hand mode is relatively large, and hence so is p(x). Using a maximum likelihood ratio test, the specification (R2) is evaluated against a single regime alternative y = y ∗ (x) + u,
(R1)
which can be thought of for the moment as being generated by a convex Solow model. The great benefit of the specifications (R1) and (R2) – as emphasized by the authors – is that long run output depends only on exogenous factors. The need to specify the precise system of endogenous variables and their interactions is circumvented.50 In conducting the test of (R1) against (R2), it is important not to include as exogenous characteristics any variable which is in fact endogenously determined. For to do so might result in conditioning on the outcomes of the underlying process which generates multiple equilibria. In the words of the authors, “Including such variables may give the impression of a unique equilibrium relationship [for the economic system] when in reality they are a function of the equilibrium being observed. Fundamental forces must be characteristics that determine a country’s economic performance but are not determined by it.” 50 Ergodicity is critical in this respect, for without it p will depend not just on x but also on the lagged values
of endogenous variables.
Ch. 5: Poverty Traps
337
In the estimation of Bloom, Canning and Sevilla, only geographic features are included in the set of exogenous variables. These include data on distance from equator, rainfall, temperature, and percentage of land area more than 100 km from the sea. For this set of variables, the likelihood ratio test rejects the single regime model (R1) in favor of the multiple equilibria model (R2). They find evidence for a high level equilibrium which does not vary with x, and a low level equilibrium which does. In particular, y1∗ (x) tends to be smaller for hot, dry, land-locked countries (and larger for those with more favorable geographical features). In addition, p(x) is larger for countries with unfavorable geographical features. In other words, the mode of the ergodic distribution around y1∗ (x) is relatively large. For these economies escape from the poverty trap is more difficult. Overall, the results of the study support the poverty trap hypothesis. They also serve to illustrate the importance of distinguishing between variables which are exogenous and those which have feedback from the system. If one conditions on “explanatory” variables which deviate significantly from fundamental forces, the likelihood of observing multiple equilibria in the map from those variables to outcomes will be lower. For example, one theme of this survey is that institutions can be an important source of multiplicity, either directly or indirectly through their interactions with the market. If institutions are endogenous, and if traps in institutions drive the disparities in cross-country incomes, then conditioning on institutions may give spurious convergence results entirely disconnected from long run outcomes generated by the system. 4.3. Model calibration One of the advantages of the methodology proposed by Bloom, Canning and Sevilla is that estimation and testing can proceed without fully specifying the underlying model. The exacting task of determining the relevant set of endogenous variables and the laws by which they interact is thereby circumvented. But there are two sides to this coin. While the results of the test suggest that poverty traps matter, they give no indication as to their source, or to the appropriate framework for formulating them as models. Graham and Temple (2004) take the opposite approach. They give the results of a numerical experiment starting from a specific poverty trap model, somewhat akin to the inertial self-reinforcement model of Section 3.4. The question they ask is whether or not the model in question has the potential to explain observed cross-country variation in per capita income for a reasonable set of parameters. We briefly outline their main findings, as well as their technique for calibration, which is of independent interest. As in Section 3.4, there is both a traditional agricultural sector and a modern sector with increasing social returns due to technical externalities. The agricultural sector has a decreasing returns technology γ
Ya = Aa La ,
γ ∈ (0, 1),
(11)
338
C. Azariadis and J. Stachurski
where Ya is output, Aa is a productivity parameter and La is labor employed in the agricultural sector. The j th firm in the modern sector has technology Ym,j = Am Lm,j Lλm ,
λ > 0,
(12)
where Ym,j is output of firm j , Am is productivity, Lm,j is labor employed by firm j , and Lm is total employment in the modern sector. The firm ignores the effect of its hiring decisions on Lm , thus setting the stage for multiplicity. We set La + Lm = L, a fixed constant, and, as usual, α := Lm /L. The relative price of the two goods is fixed in world markets and normalized to one by appropriate choice of units. Wages are determined by marginal cost pricing: γ −1 wa = γ Aa La and wm = Am Lλm . Setting these factor payments equal gives the set of equilibrium modern sector shares α as solutions to the equation (1 − α)1−γ α λ =
Aa γ Lγ −1−λ . Am
(13)
Regarding calibration, γ is a factor share, and the increasing returns parameter λ has been calculated in several econometric studies.51 Relative productivity is potentially more problematic. However, it turns out that (13) has precisely two solutions for reasonable parametric values. Since both solutions α1 and α2 satisfy (13) we have (1 − α1 )1−γ α1λ − (1 − α2 )1−γ α2λ = 0.
(14)
In which case, assuming that current observations are in equilibrium, one can take the observed share as α1 , calculate α2 as the other solution to (14), and set the poverty trap equilibrium equal to α1∗ := min{α1 , α2 }. The high productivity equilibrium is α2∗ := max{α1 , α2 }. Figure 17 illustrates this procedure for α1 = 0.1, γ = 0.7 and λ = 0.3. When α1∗ , α2∗ , γ and λ are known, a little algebra shows that the ratio of output in the high equilibrium to output in the low equilibrium can also be computed. In this way it is possible to evaluate the relative impact of the poverty trap on individual countries and the cross-country income distribution. Using this strategy and a more elaborate model (including both capital and land), Graham and Temple’s main findings are as follows. First, for reasonable parameter values some 1/4 of the 127 countries in their 1988 data set are in the poverty trap α1∗ . Second, after calculating the variance of log income across countries when all are in their high output equilibrium and comparing it to the actual variance of log income, they find that the poverty trap model can account for some 2/5 to one half of all observed variation in incomes. Overall, their study suggests that the model can explain some properties of the data, such as the difference between poor, agrarian economies and low to middle income countries. On the other hand, it cannot account for the huge differences between the very poorest and the rich industrialized countries. In the model, the largest ratios of 51 See, for example, Caballero and Lyons (2002).
Ch. 5: Poverty Traps
339
Figure 17.
low to high equilibrium production are in the region of two to three. As we saw in Section 2.1, however, actual per capita output ratios between rich and poor countries are much larger. 4.4. Microeconomic data There has also been research in recent years on poverty traps that occur at the individual or group level. For example, Jalan and Ravallion (2002) fit a microeconomic model of consumption growth with localized spillovers to farm-household panel data from rural China. Their results are consistent with empirical significance of geographical poverty traps. Other authors have studied particular trap mechanisms. For example, Bandiera and Rasul (2003) and Conley and Udry (2003) consider the effects of positive network externalities on technology adoption in Mozambique and Ghana respectively. Barrett, Bezuneh and Aboud (2001) consider the dynamic impact of credit constraints on the
340
C. Azariadis and J. Stachurski
poor in Côte d’Ivoire and Kenya. Morduch (1990) studies the effect of risk on income in India, as does Dercon (1998) for Tanzania.
5. Nonconvexities, complementarities and imperfect competition Increasing returns production under imperfect competition is a natural framework to think about multiple equilibria. Imperfect competition leads directly to externalities transmitted through the price system, because monopolists themselves, rather than Walrasian auctioneers, set prices, and presumably they do so with their own profit in mind. At the same time, their pricing and production decisions impinge on other agents. These general equilibrium effects can be a source of multiplicity. Section 5.1 illustrates this idea using the big push model of Murphy, Shleifer and Vishny (1989); a model which formalizes an earlier discussion in Rosenstein-Rodan (1943). Rosenstein-Rodan argued that modern industrial technology is freely available to poor countries, but has not been adopted because the domestic market is too small to justify the fixed costs it requires. If all sectors industrialize simultaneously, however, the market may potentially be expanded to the extent that investment in modern technology is profitable. Thus the big push model of Section 5.1 helps to clarify the potential challenges posed by coordination for the industrialization process. We shall see that the major coordination problem facing monopolists cannot be resolved by the given market structure. In this situation, the ability of a society to successfully coordinate entrepreneurial activity – and thereby realize the social benefits available in modern production technologies – will depend in general on such structures as its institutions, political organizations, the legal framework, and social and business conventions. In countries such as South Korea, the state has been very active in attempting to overcome coordination problems associated with industrialization. In Western Europe, the state was typically much less active, and the role of the private sector was correspondingly larger. For example, Da Rin and Hellmann (2002) have recently emphasized the important role played by banks in coordinating industrialization. Section 5.2 reviews their model. A theme of this survey is traps that prevent economies as a whole from adopting modern production technologies. One aspect of this transformation to modernity is the need for human capital. If investment in human capital has a high economic payoff then a skilled work-force should spontaneously arise. Put differently, if the poor are found to invest little in schooling or training then this suggests to us that returns to these investments are relatively low. Section 5.3 reviews Kremer’s (1993) matching model, where low investment in schooling sustains itself in a self-reinforcing trap. Finally, Section 5.4 gives references to notable omissions on the topic of increasing returns.
Ch. 5: Poverty Traps
341
5.1. Increasing returns and imperfect competition Murphy, Shleifer and Vishny’s (1989) formalization of Rosenstein-Rodan’s (1943) big push is something of a watershed in development economics. Their model turns on demand spillovers which create complementarities to investment. They point out that for the economy to generate multiple equilibria, it must be the case that investment simultaneously (i) increases the size of other firms’ markets, or otherwise improves the profitability of investment; and (ii) has negative net present value. This means that profits alone cannot be the direct source of the market size effects; otherwise (i) and (ii) would be contradictory. In the first model they present, higher wages in the modern sector are the channel through which demand spillovers increase market size. Although investment is not individually profitable, it raises labor income, which in turn raises the demand for other products. If the spillovers are large enough, multiple equilibria will occur. In their second model, investment in the modern technology changes the composition of aggregate demand across time. In the first period, the single monopolistic firm in each sector decides whether to invest or not. Doing so incurs a fixed cost F in the first period, and yields output ωL in the second, where ω > 1 is a parameter and L is labor input. The cost in the second period is just L, as wages are the numeraire. If, on the other hand, the monopolist chooses not to invest, production in that sector will take place in a “competitive fringe” of atomistic firms using constant returns to scale technology. For these firms, one unit of labor input yields one unit of output. The price for each unit so produced is unity. All wages and profits accrue to a representative consumer, who supplies L units of labor in both periods, and maximizes the undiscounted utility of his consumption, that is, 1 1 max ln c1 (α) dα + ln c2 (α) dα ct : [0, 1] α → ct (α) ∈ [0, ∞) 0
0
1 subject to the constraints 0 c1 p1 y1 and 0 p2 c2 y2 . Here α ∈ [0, 1] indexes the sector, ct (α) and pt (α) are consumption and price of good α at time t respectively, and yt is income (wages plus profits) at time t.52 In the first period only the competitive fringe produces, and p1 (α) = 1 for all α. In the second, monopolists face unit elastic demand curves c2 (α) = y2 /p2 (α). Given these curves and the constraints imposed by the competitive fringe, monopolists set p2 (α) = 1 for all α. Their profits are π = ay2 − F , where a := 1 − 1/ω is the mark-up. 1
52 To simplify the exposition we assume that consumers can neither save nor dissave from current income.
For the moment we also abstract from the existence of a financial sector. Firms which invest simply pay all wages in the second period at a zero rate of interest. See the original for a more explicitly general equilibrium formulation.
342
C. Azariadis and J. Stachurski
Figure 18.
Consider profitability when all entrepreneurs corresponding to sectors [0, α] decide to invest. (The number α can also be thought of as the fraction of the total number of monopolists who invest.) It turns out that for some parameter values both α = 0 and α = 1 are equilibria. To see this, consider first the case α = 0, so that y1 = y2 = L. It is not profitable for a firm acting alone to invest if π = aL − F 0. On the other hand, if α = 1, then y1 = L − F and y2 = ωL, so monopolists make positive profits when aωL − F 0. Multiple equilibria exist if these inequalities hold simultaneously. In Figure 18 multiple equilibria obtain for all L ∈ [L1 , L2 ]. As was mentioned in the introduction, coordination problems and other mechanisms that reinforce the status quo can interact with each other and magnify their individual impact. Murphy, Shleifer and Vishny (1989) provide a simple example of this in the context of the model outlined above. They point out that the coordination problem for the monopolists is compounded if industrialization requires widespread development of infrastructure and intermediate inputs, such as railways, road networks, port facilities
Ch. 5: Poverty Traps
343
and electricity grids. All of these projects will themselves need to be coordinated with industrialization. For example, suppose that n infrastructure projects must be undertaken in the first period to permit industrialization in the second. Each project has a fixed cost Rn , and operates in the second period at zero marginal cost. Leaving aside the issue of how the spoils of industrialization will be divided among the owners of the projects and the continuum of monopolists, it is clear that industrialization has the potential to be profitable for all only when aωL − F , the profits of the monopolists when α = 1, exceed total infrastructure costs ni=1 Ri . If the condition aL − F 0 continues to hold, however, individual monopolists investing alone will be certain to lose money. Realizing this, investors in infrastructure face extrinsic uncertainty as to whether or not industrialization will actually take place. Given their subjective evaluation, they may choose not to start their infrastructure projects. In turn, the monopolists are aware that investors in infrastructure face uncertainty, and may themselves refrain from starting projects. This makes monopolists even more uncertain as to whether or not the conditions for successful industrialization will eventuate. The fixed point of this infinite regression of beliefs may well be inaction. In either case, the addition of more actors adds to the difficulty of achieving coordination. 5.2. The financial sector and coordination As Da Rin and Hellmann (2002) have recently emphasized, one candidate within the private sector for successfully coordinating a big push type industrialization is the banks. Banks are the source of entrepreneurs’ funds, and shape the terms and conditions under which capital may be raised. In addition, banks interact directly with many entrepreneurs. Finally, banks can potentially profit from coordinating industrialization if their market power is large. Da Rin and Hellmann find that the structure and legal framework of the banking sector are important determinants of its ability to coordinate successful industrialization. To illustrate their ideas, consider again the big push model of Section 5.1. In order to make matters a little easier, let us simply define the second period return of monopolists (entrepreneurs) to be f (α), where α is the fraction of entrepreneurs who decide to set up firms and the function f : [0, 1] → R is strictly increasing. As before, there is a fixed cost F to be paid in the first period, which we set equal to 1. The future is not discounted. It is convenient to think of the number of entrepreneurs as some large but finite number N .53 In addition to these N entrepreneurs, there is now a financial sector, members of whom are referred to as either “banks” or “investors”. There are B ∈ N banks, the first B − 1 of which have an intermediation cost of r per unit of investment. The last 53 In particular, entrepreneurs do not take into account their influence on α when evaluating whether to set
up firms or not.
344
C. Azariadis and J. Stachurski
bank has an intermediation cost of zero, but can lend to only N firms. The number can be thought of as a measure of the last bank’s market power. The equilibrium lending rate at which firms borrow in the first period is determined by the interaction of the monopolists and the banks. In the first stage of the game, each bank b offers a schedule of interest rates to the N firms. This strategy will be written as σb := {inb : 1 n N}. The collection of these strategies across banks will be written as σ := {σb : 1 b B}. Let Σ be the set of all such σ . In the second stage, each entrepreneur either rejects all offers and does not set up the firm, or selects the minimum interest rate, pays the fixed cost and enters the market. In what follows we write mn (σ ) to mean minb inb , the minimum interest rate offered to firm n in σ . If a fraction α accepts contracts then firm n makes profits π α, mn (σ ) = f (α) − 1 + mn (σ ) . (15) For bank b < B, profits are given by Πb (σb ) =
N b in − r 1{firm n accepts},
(16)
n=1
where here and elsewhere 1{Q} is equal to one when the statement Q is true and zero otherwise. For b = B, profits are Πb (σb ) =
N
inb 1{firm n accepts}.
(17)
n=1
In equilibrium, banks never offer interest rates strictly greater than r, because should they do so other banks will always undercut them. As a result, we can and do assume in all of what follows that mn (σ ) r for all n. Also, to make matters interesting, we assume that f (0) < 1 + r < f (1), or, equivalently, π(0, r) < 0 < π(1, r).
(18)
Firms’ actions will depend on their beliefs – in particular, on what fraction α of the N firms they believe will enter. Clearly beliefs will be contingent on the set of contracts offered by banks. Thus a belief for firm n is a map αne from Σ into [0, 1]. Given this belief, firm n enters if and only if π αne (σ ), mn (σ ) 0. (19) Given σ , the set of self-supporting equilibria for the second stage subgame is N 1 Ω(σ ) := α ∈ [0, 1]: 1 π α, mn (σ ) 0 = α . N
(20)
n=1
In other words, α ∈ Ω(σ ) if, given the set of offers σ and the belief on the part of all firms that the fraction of firms entering will be α, exactly α × 100% of firms find it optimal to enter.
Ch. 5: Poverty Traps
345
Beliefs are required to be consistent in the sense that αne (σ ) ∈ Ω(σ ) for all σ and all n. Beliefs are called optimistic if αne = α opt for all n, where α opt (σ ) := max Ω(σ ) for all σ ∈ Σ. In other words, all agents believe that as many firms will enter as are consistent with offer σ , and this is true for every σ ∈ Σ. Beliefs are defined to be pessimistic if the opposite is true; that is, if αne = α pes for all n, where α pes (σ ) := min Ω(σ ) for all σ ∈ Σ. Da Rin and Hellmann first observe that if = 0, then the outcome of the game will be determined by beliefs. In particular, if beliefs are pessimistic, then the low equilibrium α = 0 will obtain. If beliefs are optimistic, then the high equilibrium α = 1 will obtain. The interpretation is that when = 0, so that the market for financial services is entirely competitive (in the sense of Bertrand competition with identical unit costs described above), the existence of the financial sector will not alter the primary role of beliefs in determining whether industrialization will take place. Let us verify this observation in the case of pessimistic beliefs. To do so, it is sufficient to show that if σ ∈ Σ is optimal, then 0 ∈ Ω(σ ). The reason is that if 0 ∈ Ω(σ ), then by (20) we have π(0, mn (σ )) < 0 for all n. Also, beliefs are pessimistic, so αne (σ ) = min Ω(σ ) = 0. In this case no firms enter by (19). To see that 0 ∈ Ω(σ ) for all optimal σ , suppose to the contrary that σ ∈ Σ is optimal, but 0 ∈ / Ω(σ ). Then π(0, mk (σ )) 0 for some k, in which case (18) implies that mk (σ ) < r. Because firms only accept contracts at rates less than r (that is, mn (σ ) r for all n), it follows from (16) that the bank which lent to k looses money, and σ is not optimal. The intuition is that no bank has market power, and cannot recoup losses sustained when encouraging firms to enter by offering low interest rates. More interesting is the case where the last bank B has market power. With sufficient market power, B will induce industrialization (the high equilibrium where α = 1) even when beliefs are pessimistic: P ROPOSITION 5.1 (Da Rin and Hellmann). Suppose beliefs are pessimistic. In this case, there exists an α¯ ∈ [0, 1] depending on r and f such that industrialization will occur whenever , the market power of B, satisfies /N α. ¯ The result shows that rather than relying on spontaneous coordination of beliefs, financial intermediaries may instead be the source of coordination. The key intuition is that a financial intermediary may have a profit motive for inducing industrialization. But to achieve this, two things are necessary: size and market power. Size (as captured by ) is necessary to induce a critical mass of entrepreneurs to invest. Market power (as captured by the cost advantage r) is necessary to recoup the costs of mobilizing that critical mass. We sketch Da Rin and Hellmann’s proof in Appendix A. Until now we have considered only the possibility that the banks offer pure debt contracts. Da Rin and Hellmann also study the case where the banks may hold equity as well (i.e., universal banking). They show that in this case the threshold level at which the lead bank B has sufficient market power to mobilize the critical mass is lower. Industrialization is unambiguously more likely to occur. The reason is that equity permits B
346
C. Azariadis and J. Stachurski
to partake in the ex post profits of the critical mass, who benefit from low interest rates on one hand and complete entry (α = 1) on the other. With a lower cost of mobilizing firms, B requires less market power to recoup these losses. In Da Rin and Hellmann’s words, Our model provides a rationale for why a bank may want to hold equity that has nothing to do with the standard reasons of providing incentives for monitoring. Instead, equity allows a bank to participate in the gains that it creates when inducing a higher equilibrium. In summary, the theory suggests that large universal banks with a high degree of market power can play a central role in the process of industrialization. This theory is consistent with the evidence from countries such as Belgium, Germany and Italy, where a few oligopolist banks with strong market positions played a pivotal role. Some were pioneers of universal banking, and many directly coordinated activity across sectors by participation in management. The theory may also explain why other countries, such as Russia, failed to achieve significant industrialization in the 19th Century. There banks were small and dispersed, their market power severely restricted by the state. 5.3. Matching The next model we consider is due to Kremer (1993), and has the following features. A production process consists of n distinct tasks, organized within a firm. For our purposes n can be regarded as exogenous. The tasks are undertaken by n different workers, all of whom have their own given skill level hi ∈ [0, 1]. Here the skill level will be thought of as the probability that the worker performs his or her task successfully. We imagine that if one worker fails in their task the entire process is ruined and output is zero. If all are successful, the outcome of the process is n units of the product.54 That is, y=n
n
1{worker i successful},
P{worker i successful} = hi ,
(21)
i=1
where as before 1{Q} = 1 if the statement Q is true and zero otherwise. All of the success probabilities are independent, so that E(y) = n i hi . Consider an economy with a unit mass of workers. The distribution of skills across workers is endogenous, and will be discussed at length below. Kremer’s first point is that in equilibrium, firms will match workers of equal skill together to perform the process. The intuition is that (i) firms will not wish to pair a work-force of otherwise skilled employees with one relatively unskilled worker, who may ruin the whole process; and (ii) firms with a skilled work-force will be able to bid more for skilled workers, because the marginal value of increasing the last worker’s skill is increasing in the skill of the 54 Assuming one unit might seem more natural than n, but the latter turns out to be more convenient.
Ch. 5: Poverty Traps
347
other workers. Thus, for each firm, E(y) = nhn ,
h the firm’s common level of worker skill.
(22)
The first thing to notice about this technology is that the expected marginal return to skill is increasing. As a result, small differences in skill can have relatively large effects on output. This may go some way to explaining the extraordinarily large wage differentials between countries. Moreover, for economies with such technology, positive feedback dynamics of the kind considered in Section 3.3 may result, even if the technology for creating human capital is concave. Another channel for positive feedbacks occurs when matching is imperfect, perhaps because it is costly or the population is finite. Exact matches may not be possible. In that case, there are potentially returns to agglomeration: Skilled people clustering together will decrease the cost of matching, and increase the likelihood of good matches. Also, an initial distribution of skills will tend to persist, because workers will choose skills so as to be where the distribution is thickest. This maximizes their chances of finding good matches. But this is self-reinforcing: Their choices perpetuate the current shape of the distribution. There is yet another channel that Kremer suggests may lead to multiple equilibrium distributions of skill. This is the situation where skill levels are imperfectly observed. We present a simple (and rather extremist) version of Kremer’s model. In the first period, workers decide whether to undertake “schooling” or not. This education involves a common cost c ∈ (0, 1). In the second, firms match workers, produce, and pay out wages. Both goods and labor markets are competitive, and total wages exhaust revenue. Specifically, it is assumed that each worker’s wage w is (1/n)th of firm’s output. Not all of those who undertake schooling become skilled. We assume that the educated receive a skill level h = 1 with probability p > 1/2 and h = 0 with probability 1 − p. Those who do not undertake schooling have the skill level h = 0. Further, h is not observable, even for workers. Instead, all workers take a test, which indicates their true skill with probability p and the reverse with probability 1 − p.55 That is, h with probability p; t := test score = (23) 1 − h with probability 1 − p. Firms then match workers according to the test score t rather than h. Let α ∈ [0, 1] denote the fraction of workers who choose to undertake schooling. We will show that for certain values of the parameters p and c, both α = 0 and α = 1 are equilibria. In doing so, we assume that p is known to all. Also, workers and firms are risk neutral. Consider first the case where α = 0. If the worker undertakes schooling, then, re gardless of his skill and test score, his expected wage is (1/n)th of n i hi , where his co-workers are drawn from a pool in which the skilled workers have measure zero. That 55 We are using the same p as before just to simplify notation.
348
C. Azariadis and J. Stachurski
is, P{hi = 0} = 1. It follows that expected output and wage are zero. Since c > 0, it is optimal to avoid schooling.56 Now consider the agent’s problem when α = 1. In the second period, the agent will be matched with other workers having the same test score. In either case, computing expected wages is a signal extraction problem. First, using the fact that agents in the pool of potential co-workers have chosen schooling with probability one, the agent can calculate probable skills of a co-worker chosen at random from the population, given their test score: P{h = 1 | t = 1} =
P{h = 1 and t = 1} p2 =: θp , = 2 P{t = 1} p + (1 − p)2
(24)
P{h = 1 | t = 0} =
p(1 − p) 1 P{h = 1 and t = 0} = = . P{t = 0} p(1 − p) + p(1 − p) 2
(25)
and,
The worker can use these probabilities to compute expected output and hence wages given the different outcomes of his own test score. In particular, E(w | t = 1) = θpn and E(w | t = 0) = (1/2)n . It follows that the expected return to schooling for the agent is E(w | schooling) = E(w | t = 0)P{t = 0} + E(w | t = 1)P{t = 1} 1 = n (1 − p) + θpn p. 2 Conversely, E(w | no schooling) =
1 2n p
+ θpn (1 − p). Schooling is optimal if
c < E(w | schooling) − E(w | no schooling) = (2p − 1) θpn − (1/2)n := c∗ (p). It is easy to see that c∗ (p) > 0 whenever p > 1/2, which is true by assumption. As a result, schooling will be optimal for some sufficiently small c, and α = 1 is an equilibrium too.57 What are the sources of multiple equilibria in the model? The first is pecuniary externalities in the labor market: When more agents become educated, the probability that the marginal worker can successful match with a skilled co-worker increases. In turn, this increases the returns to education.58 Second, there is imperfect information: Skilled workers cannot readily match with other skilled workers. Instead, matching is
56 On the other hand, if skills are perfectly observable, workers who acquire skills will be matched with n
workers from the measure zero set of agents having h = 1. In that case w = 1. Since c < 1 it is optimal to choose schooling, and α = 0 is not an equilibrium. The same logic works for any α < 1. 57 It may seem that if p = 1 and observation is perfect, then E(w | schooling) − E(w | no schooling) should be zero, so that no multiple equilibria are possible. But under this assumption the above derivation of c∗ (p) is not valid, because we would be conditioning on sets with probability zero. 58 In fact the expected wage is increased for all, but those who become skilled benefit more.
Ch. 5: Poverty Traps
349
probabilistic, and depends on the overall distribution of skills. Finally, the increasing expected marginal reward for skill inherent in the production function means that the wage spillovers from the decisions of other agents are potentially large. Another important model of human capital investment with multiple equilibria is Acemoglu (1997). He shows how labor market frictions can induce a situation where technology adoption is restricted by a lack of appropriately skilled workers. Low adoption in turn reduces the expected return to training, further exacerbating the scarcity of workers who are trained. In other words, poor technology adoption and low capital investment are self-reinforcing, because they cause the very shortage of skilled workers necessary to make such investments profitable. 5.4. Other studies of increasing returns Young’s (1928) famous paper on increasing returns notes that not only does the degree of specialization depend on the size of the market, but the size of the market also depends on the degree of specialization. In other words, there are efficiency gains from greater division of labor, primarily due to application of machines. Greater specialization increases productivity, which then expands the market, leading back into more specialization, and so on. As a result, there are complementarities in investment. These complementarities can be the source of poverty traps. A detailed discussion of this process is omitted from the present survey, but only because excellent surveys already exist. See in particular Matsuyama (1995) and Matsuyama (1997). Other references include Matsuyama and Ciccone (1996), Rodríguez-Clare (1996) and Rodrik (1996). Increasing returns are also associated with geographical agglomeration. Starrett (1978) points out that agglomerations cannot form as the equilibria of perfectly competitive economies set in a homogeneous space. Thus all agglomerations must be caused either by exogenous geographical features or by some market imperfection. An obvious candidate is increasing returns. (It is difficult to see what geographical features could explain the extent of concentration witnessed in places such as Tokyo or Hong Kong.) This survey does not treat geography and its possible connections with poverty traps in much detail. Interested readers might start with the review of Ottaviano and Thisse (2004).59 Another source of complementarities partly related to geography is positive network externalities in technology adoption. These are often thought to arise from social learning: Local experience with a technology allows the cost of adoption to decrease as the number of adopters in some network gets larger. As well as information spillovers, more adopters of a given technology may lead to the growth of local supply networks for intermediate inputs, repairs and servicing, skilled labor and so on. See, for example, Beath, Katsoulacos and Ulph (1995), Bandiera and Rasul (2003), Conley and Udry (2003), and Baker (2004). 59 See also Limao and Venables (2001) or Redding and Venables (2004) for the empirics of geography and
international income variation.
350
C. Azariadis and J. Stachurski
Finally, an area that we have not treated substantially in this survey is optimal growth under nonconvexities, as opposed to the fixed savings rate model considered in Section 3.3. In other words, how do economies evolve when (i) agents choose investment optimally by dynamic programming, given a set of intertemporal preferences; and (ii) the aggregate production function is nonconvex? There are two main cases. One is that increasing returns are taken to be external, perhaps as a feedback from aggregate capital stock to the productivity residual, and agents perceive the aggregate production function to be convex. In this case there is a subtle issue: In order to optimize, agents must have a belief about how the productivity residual evolves. This may or may not coincide with its actual evolution as a result of their choices. An equilibrium transition rule is a specification of savings and investment behavior such that (a) agents choose this rule given their beliefs; and (b) those choices cause aggregate outcomes to meet their expectations. Existence of such an equilibrium is far from assured. See Mirman, Morand and Reffett (2004) and references therein. Dynamics are still actively being investigated. The second case is where increasing returns are internal, and agents perceive aggregate production possibilities exactly as they are. These models generate similar poverty traps as were found for fixed savings rates in Section 3.3. The literature is large. An early investigation is Skiba (1978). See also Dechert and Nishimura (1983), who consider a per capita production function k → f (k) which is convex over a lower region of the state space (capital per worker), and concave over the remainder; and Amir, Mirman and Perkins (1991), who study the same problem using lattice programming. Majumdar, Mitra and Nyarko (1989) study optimal growth for stochastic nonconvex models, as do Nishimura and Stachurski (2004). Dimaria and Le Van (2002) analyze the dynamics of deterministic models with R&D and corruption.60
6. Credit markets, insurance and risk In terms of informational requirements necessary for efficient free market operation and low transaction costs, one of the most problematic of all markets is the intertemporal trade in funds. Here information is usually asymmetric, and lenders face the risk of both voluntary and involuntary default [Kehoe and Levine (1993)]. Voluntary default is strategic default by borrowers who judge the expected rewards of repayment to be lower than those of not repaying the loan. Involuntary default occurs when ex post returns are insufficient to cover total loans. Facing these risks, a standard response of lenders is to make use of collateral [Kiyotaki and Moore (1997)]. But the poor lack collateral almost by definition; as a result they are credit constrained. Credit constraints in turn restrict participation by the
60 One should be cautious about interpreting these nonconvex models as aggregative studies of development.
The Second Fundamental Welfare Theorem does not apply, so decentralization is problematic.
Ch. 5: Poverty Traps
351
poor in activities with substantial set up costs, as well as those needing large amounts of working capital. For the poor, then, the range of feasible income-generating activities is reduced. Thus, the vicious circle of poverty: Income determines wealth and low wealth restricts collateral. This trap is discussed in Section 6.1.61 The market for insurance is similar to the market for credit, in that information is asymmetric and transaction costs are high. This can lead to poverty traps in several ways. In Section 6.2, we study a model where poor agents, lacking access to insurance or credit, choose low risk strategies at the cost of low mean income. These choices reinforce their poverty. In Section 6.3 we review Matsuyama’s (2004) world economy model, where all countries must compete for funds in a global financial market. On one hand, diminishing returns imply that rewards to investment in the poor countries are large. High returns attract funds and investment, and high investment provides a force for convergence. On the other hand, credit markets are imperfect, and rich countries have more collateral. This puts them in a strong position vis-a-vis the poor when competing for capital. The inability of the poor to guarantee returns with collateral is a force for divergence. 6.1. Credit markets and human capital Consider an economy producing only one good and facing a risk free world interest rate of zero. Agents live for one period. Each has one and only one child. From their parent, the child receives a bequest x. At the beginning of life, each agent chooses between two occupations. The first is to work using a constant returns technology Y = wL, ¯ where Y is output, L is total labor input in this sector, and w¯ is a productivity parameter. The agent supplies all of his or her labor endowment t , and we define wt := w ¯ t as the return to this choice of occupation. We admit the possibility that t varies stochastically, so wt may be random. Alternatively, the agent may set up a project at cost F . The gross payoff from the project is equal to Qt . Agents with wealth xt < F may borrow to cover the costs of the project beyond which they are able to self-finance. They face interest rate i > 0, where the excess of the borrowing rate over the risk free rate reflects a credit market imperfection. In this case we have in mind costs imposed on lenders due to the need for supervision and contract enforcement [cf., e.g., Galor and Zeira (1993, p. 39)]. These costs are then passed on to the borrower. The two stochastic productivity parameters wt and Qt are draws from joint distribution ϕ. We assume that Et = 1, and that Ewt = w¯ < EQt − F . Thus, the net return to setting up the project is higher on average than the wage. However, the agent may still choose to work at wage rate wt if his or her income is relatively low. The reason is that
61 See also Tsiddon (1992) for a poverty trap model connected to the market for credit. In his model, asym-
metric information leads to a moral hazard problem, which restricts the ability of investors to raise money. The market solution involves quantity constraints on loans, the severity of which depends on the level of income.
352
C. Azariadis and J. Stachurski
for the poor setting up a project requires finance at the borrowing rate i > 0, which may offset the differential return between the two occupations. Consider the employment decisions and wealth dynamics for each dynasty. Omitting time subscripts, an agent with bequest x has x + w if do not set up project; y := lifetime income =
(x − F )(1 + i) + Q (x − F ) + Q
if set up project, x < F ; if set up project, x F.
Preferences are given by u(c, b) = (1 − θ ) ln c + θ ln b, where θ ∈ (0, 1) is a parameter, c is consumption and b is bequest to the child. As a result, each agent bequeaths a fraction θ of y; the remainder is consumed. Indirect utility is v(y) = γ + δ ln y, where γ , δ > 0 are constants. To abstract temporarily from the issue of risk aversion let us suppose that each agent can observe his or her idiosyncratic shocks (w, Q) prior to choosing a field of employment. As a result, agents with x F will choose to set up projects iff Q − F w. Agents with x < F will choose the same iff (x − F )(1 + i) + Q x + w; in other words, iff w − Q + F (1 + i) . i It follows that dynamics for each dynasty’s wealth in this economy are given by the transition rule x t + wt if xt xˆt ; xt+1 = St (xt ); St (xt ) = θ × (xt − F )(1 + i) + Qt if xt ∈ (xˆt , F ); if xt F. xt − F + Qt Figure 19 illustrates a transition rule S and hence the dynamics of this economy when the two rates of return are constant and equal to their means.62 For this particular parameterization there are multiple equilibria. Agents with initial wealth less than the critical value xb will converge to the lower attractor, while those with greater wealth will converge to the high attractor. Given any initial distribution ψ0 of wealth x in the economy, the fraction of agents converging to the lower attractor will be 0 b dψ0 . If this fraction is large, average long run income in the economy will be small. A more realistic picture can be obtained if the productivity parameters are permitted to vary stochastically around their means. This will allow at least some degree of income mobility – perhaps very small – which we tend to observe over time in almost all societies. To this end, suppose that for each agent and at each point in time the parameters wt and Qt are drawn independently across time and agents from a bivariate lognormal distribution. In this case the transition law is itself random, and varies for each agent at each point in time. Figure 20 shows a simulated sequence of transition rules facing a given agent starting at t = 1. At t = 2 a negative shock to the project return Q causes the high level attractor x xˆ :=
62 The parameters here are set to θ = 0.7, w = 0.06, Q = 1.05, i = 2 and F = 0.65.
Ch. 5: Poverty Traps
353
Figure 19.
to disappear. A series of such negative shocks would cause a rich dynasty to loose its wealth. In this case, however, the shocks are iid and such an outcome is unlikely. It turns out that the time 3 shocks are strongly positive. If the number of agents is large, then the sequence of cross-sectional distributions for wealth over time can be identified with the sequence of marginal probability laws (ψt )t0 generated by the Markov process xt+1 = St (xt ). It is not difficult to prove that this Markov process is ergodic. The intuition and the dynamics are more or less the same as for the nonconvex growth model of Section 3.3.63 We postpone further details on dynamics until the next section, which treats another version of the same model. There are several interpretations of the two sector story with fixed costs described above. One is to take the notion of a project or business literally, in which case F is 63 As we discussed at length in that section, it would be a mistake to claim that this ergodicity result in some
way overturns the poverty trap found in the deterministic version.
354
C. Azariadis and J. Stachurski
Figure 20.
the cost of set up and working capital which must be paid up before the return is received. Alternatively F might be the cost of schooling, and Q is the payoff to working for skilled individuals.64 As emphasized by Loury (1981) and others, human capital is particularly problematic for collateral-backed financing, because assets produced by investment in human capital cannot easily be bonded over to cover the risk of default.
64 For these and related stories see Ray (1990), Ray and Streufert (1993), Banerjee and Newman (1993),
Galor and Zeira (1993), Ljungqvist (1993), Freeman (1996), Quah (1996), Aghion and Bolton (1997), Piketty (1997), Matsuyama (2000), Mookherjee and Ray (2003) and Banerjee (2003). Yet another possible interpretation of the model is that F is the cost of moving from a rural to an urban area in order to find work. In the presence of imperfect capital markets, such costs – interpreted broadly to include any extra payments incurred when switching to the urban sector – may help to explain the large and growing differentials between urban and rural incomes in some modernizing countries.
Ch. 5: Poverty Traps
355
Whatever the precise interpretation, the “project” represents an opportunity for the poor to lift themselves out of poverty, while the fixed cost F and the credit market imperfection captured here by i constitute a barrier to taking it. Microeconometric studies suggests that the effects of this phenomenon are substantial. For example, Barrett, Bezuneh and Aboud (2001) analyze the effects of a large devaluation of the local currency that occurred in Côte d’Ivoire in 1994 on rural households. They find that “A macro policy shock like an exchange rate devaluation seems to create real income opportunities in the rural sector. But the chronically poor are structurally impeded from seizing these opportunities due to poor endowments and liquidity constraints that restrict their capacity to overcome the bad starting hand they have been dealt.” [Barrett et al. (2001, p. 12)] The same authors also study a local policy shock associated with food aid distribution in Keyna. According to this study, “The wealthy are able to access higher-return niches in the non-farm sector, increasing their wealth and reinforcing their superior access to strategies offering better returns. Those with weaker endowments ex ante are, by contrast, unable to surmount liquidity barriers to entry into or expansion of skilled nonfarm activities and so remain trapped in lower return . . . livelihood strategies.” [Barrett et al. (2001, p. 15)] 6.2. Risk For the poor another possible source of historical self-reinforcement is risk. In the absence of well-functioning insurance and credit markets, the poor find ways to mitigate adverse shocks and to smooth out their consumption. One way to limit exposure is to pass up opportunities which might seem on balance profitable but are thought to be too risky. Another strategy is to diversify activities; and yet another is to keep relatively large amounts of assets in easily disposable form, rather than investing in ventures where mean return is high. All of these responses of the poor to risk have in common the fact that they tend to lower mean income and reinforce long run poverty. A simple variation of the model from the previous section illustrates these ideas.65 Let the framework of the problem be the same, but current shocks are no longer assumed to be previsible. In other words, each agent must decide his or her career path before observing the shocks wt and Qt which determine individual returns in each sector. Given that preferences are risk averse (indirect utility is v(y) = γ + δ ln y), the agent makes these decisions as a function not only of mean return but of the whole joint distribution. Regarding this distribution, we assume that both shocks are lognormal and may be correlated. Lenders also cannot observe these variables at the start of time t, and hence the borrowing rate i = i(x) reflects the risk of default, which in turn depends on the wealth x of the agent. In particular, default occurs when Qt is less than the debtor’s total obligations 65 What follows is loosely based on Banerjee (2003).
356
C. Azariadis and J. Stachurski
(F − x)(1 + i(x)). In that case the debtor pays back what he or she is able. Lifetime income is therefore x + w if do not set up project; y = max{0, (x − F )(1 + i(x)) + Q} if set up project, x < F ; (x − F ) + Q if set up project, x F. It turns out that in our very simplistic environment agents will never borrow, because when shocks are lognormal agents with x < F who borrow will have P{y = 0} > 0, in which case Ev(y) = −∞. (If x F agents may still choose to work for a wage, depending on the precise joint distribution.) The result that agents never borrow is clearly unrealistic. For more sophisticated versions of this model with similar dynamics see Banerjee (2003) or Checchi and García-Peñalosa (2004). Because agents never borrow, the dynamics for the economy are just / D}, xt+1 = θ (xt + wt ) · 1{xt ∈ D} + θ (xt − F + Qt ) · 1{xt ∈ where D := {x: E v(x + wt ) E v(x − F + Qt )}. (As before, 1 is the indicator function.) The stochastic kernel Γ for this process can be calculated separately for the two cases x ∈ D and x ∈ / D using the same change-of-variable technique employed in Section 3.1. The calculation gives x − θx 1 x − θ (x − F ) 1 · 1{x ∈ D} + ϕQ · 1{x ∈ / D}, Γ x, x = ϕw θ θ θ θ where ϕw and ϕQ are the marginal densities of w and Q respectively. A two-dimensional plot of the kernel is given in Figure 21, where the parameters are F = 1, θ = 0.45, ln w ∼ N (0.1, 1), and ln Q ∼ N (1.4, 0.2). The dark unbroken line is the 45◦ line. Lighter areas indicate greater elevation, in this case associated with a collection of probability mass. For the parameters chosen, agents work precisely when x < F , and set up projects when x F (so that D = [0, F ]), despite the fact that mean returns to the project are higher than those of working. The concentration of probability mass along the 45◦ line in the region D = [0, F ] implies that poverty will be strongly self-reinforcing. Nevertheless, lognormal shocks give poor individuals a non-zero probability of becoming rich at every transition; and the rich can eventually become poor, although it might take a sequence of negative shocks. The rate of mixing depends on the parameters that make up the law of motion and the variance of the shock. Usually some small degree of mixing is a more natural assumption than none. The mixing causes the corresponding Markov chain to be ergodic. This is the case regardless of how small the tails of the shocks are made.66 For more details on ergodicity see Appendix A. To summarize, the poor are not wealthy enough to self-insure, and as a result choose income streams that minimize risk at the expense of mean earnings. The effect is to
66 But not necessarily so if the shocks have bounded support.
Ch. 5: Poverty Traps
357
Figure 21.
reinforce poverty. A number of country studies provide evidence of this behavior.67 Dercon (2003) finds that the effects on mean income are substantial. In a review of the theoretical and empirical literature, he estimates that incomes of the poor could be 25– 50% higher on average if they had the same protection against shocks that the rich had as a result of their wealth [Dercon (2003, p. 14)]. A more sophisticated model of the relationship between risk and development is Acemoglu and Zilibotti (1997). In their study, indivisibilities in technology imply that diversification possibilities are tied to income. An increase in investment raises output, which then improves the extent of diversification. Since agents are risk averse, greater diversification encourages more investment. In the decentralized outcome investment is too small, because agents do not take into account the effect of their investment on the diversification opportunities of others. 67 See, for example, Morduch (1990) and Dercon (1998).
358
C. Azariadis and J. Stachurski
6.3. Credit constraints and endogenous inequality Next we consider a world economy model with credit market imperfections due to Matsuyama (2004). For an individual country, the formulation of the problem is as follows. A unit mass of agents live for two periods each, supplying one unit of labor in the first period of life and consuming all their wealth in the second. Per capita output of the consumption good is given by yt = f (kt )ξt , where f is a standard concave production function, kt is the capital stock and (ξt )t0 is a noise process. Once the current shock ξt is realized production then takes place. Factor markets are competitive, so that labor and capital receive payments wt = [f (kt ) − kt f (kt )]ξt =: w(kt , ξt ) and t = f (kt )ξt respectively. Current wages wt are invested by young agents to finance consumption when old. Funds can be invested in a competitive capital market at gross interest rate Rt+1 , or in a project which transforms one unit of the final good into Q units of the capital good at the start of next period. It is assumed that projects are discrete and indivisible: Each agent can run one and only one project.68 They will need to borrow 1 − wt , the excess cost of the project over wages. Our agents are risk neutral. Time t information is summarized by the information set Ft , and we normalize E[ξt+1 | Ft ] = 1. In the absence of borrowing constraints, agents choose to start a project if E[t+1 Q − Rt+1 (1 − wt ) | Ft ] E[Rt+1 wt | Ft ]. This is equivalent to E[Rt+1 | Ft ] E[t+1 Q | Ft ].
(26)
However, it is assumed that borrowers can credibly commit to repay only a fraction λ of revenue t+1 Q. Thus λ ∈ [0, 1] parameterizes the degree of credit market imperfection faced by borrowers in this economy. As a result, agents can start a project only when E[λt+1 Q | Ft ] exceeds E[Rt+1 (1 − wt ) | Ft ], the cost of funds beyond those which the agent can self-finance. In other words, when wt = w(kt , ξt ) < 1, we must have E[Rt+1 | Ft ] Λ(kt , ξt )E[t+1 Q | Ft ],
(27)
where Λ(kt , ξt ) := λ/(1 − wt ) Given the profitability constraint (26), the borrowing constraint (27) binds only when Λ(kt , ξt ) < 1.69 In the case of autarky it turns out that adjustment of the domestic interest rate can always equilibrate domestic savings and domestic investment. Since each generation of agents has unit mass, total domestic savings is just wt . If wt 1, then all agents run projects and total output of the capital good is Q. If wt < 1, then wt is equal to the fraction of agents who can start projects. Output of the capital good is wt Q. Assuming that capital depreciates totally in each period, we get kt+1 = min{w(kt , ξt )Q, Q}. If, 68 Put differently, we imagine that output is Q units of capital good for all investment levels greater than or
equal to one. See the original model for a more general technology. 69 Of course if w 1 then all agents can self-finance and the borrowing constraint never binds. t
Ch. 5: Poverty Traps
359
Figure 22.
for example, technology in the final good sector is Cobb–Douglas, so that f (k) = Ak α , where α < 1, then w(kt , ξt ) = (1 − α)Ak α ξt . For ξt ≡ 1 there is a unique and globally stable steady state k ∗ . A more interesting case for us is the small open economy. Here a world interest rate of R is treated as fixed and given. The final good is tradable, so international borrowing and lending are allowed. However, the project must be run in the home country (no foreign direct investment) and factors of production are nontradable. In the open economy setting there is a perfectly elastic supply of funds at the world interest rate R. The effective demand for funds on the part of domestic projects is determined by (26) and (27). The right-hand side of (26) is the expected marginal product of capital in this sector, E[t+1 Q | Ft ]. Since E[ξt+1 |Ft ] = 1 we have E[t+1 Q | Ft ] = f (kt+1 )Q. Absent borrowing constraints, investment adjusts to equalize f (kt+1 )Q with R. Figure 22 shows the intersection of the curve k → f (k)Q with the horizontal supply curve R at Φ(R/Q), where Φ is the inverse function of f .
360
C. Azariadis and J. Stachurski
Figure 23.
As the figure is drawn, however, Λ(kt , ξt ) < 1, perhaps because the capital stock is small, or because of an adverse productivity shock. As a result, the borrowing constraint is binding, and next period’s capital stock kt+1 is given by the intersection of the effective demand curve k → Λ(kt , ξt )f (k)Q and the supply curve R. Assuming that Φ(R/Q) < Q as drawn in the figure, the law of motion for the capital stock is kt+1 = Ψ (kt , ξt ), where Φ[R/Λ(k, ξ )Q] if w(k, ξ ) < 1 − λ; Ψ (k, ξ ) := (28) Φ(R/Q) if w(k, ξ ) 1 − λ. For w(kt , ξt ) < 1 − λ we have Λ(kt , ξt ) < 1 and the borrowing constraint binds. Domestic investment is insufficient to attain the unconstrained equilibrium Φ(R/Q). In this region of the state space, the law of motion k → Ψ (k, ξ ) is increasing in k. Behind this increase lies a credit multiplier effect: Greater domestic investment increases
Ch. 5: Poverty Traps
361
Figure 24.
collateral, which alleviates the borrowing constraint. This in turn permits more domestic investment, which increases collateral, and so on. Individual agents do not take into account the effect of their actions on the borrowing constraint. Figure 23 shows the law of motion when ξt ≡ 1. As drawn, there is a poverty trap at kL and another attractor at Φ(R/Q). Countries with kt > kU tend to Φ(R/Q), while those with kt < kU tend to kL . Figure 24 shows stochastic dynamics by superimposing the first 50 laws of motion from a simulation on the 45◦ diagram. The shocks (ξt )t0 are independent and identically distributed.70 Notice that for particularly good shocks the lower attractor kL disappears, while for particularly bad shocks the higher attractor at Φ(R/Q) vanishes. 70 The production function is f (k) = k α . The shock is lognormal. The parameters are α = 0.59, Q = 2.4,
λ = 0.40, R = 1 and ln ξ ∼ N (0.01, 0.08).
362
C. Azariadis and J. Stachurski
Figure 25.
Figure 25 shows a simulated time series for the same parameters as Figure 24 over 500 periods. At around t = 300 the economy begins a transition to the higher attractor Φ(R/Q). Subsequent fluctuations away from this equilibrium are due to shocks so negative that Φ(R/Q) ceases to be an attractor (see Figure 24). The story does not end here. What is particularly interesting about Matsuyama’s study is his analysis of symmetry-breaking. He shows the following for a large range of parameter values: For a world economy consisting of a continuum of such countries, the deterministic steady state for autarky, which is k ∗ defined by k ∗ = w(k ∗ , 1)Q, is precisely kU , the unstable steady state for each country under open international financial markets and a world interest rate that has adjusted to equate world savings and investment. Figure 26 illustrates the situation. Thus, the symmetric steady state after liberalization, where each country has capital stock k ∗ , is unstable and cannot be maintained under any perturbation. The reason is that countries which suffer from bad (resp., good) shocks are weakened (resp., strengthened)
Ch. 5: Poverty Traps
363
Figure 26.
in terms of their ability to guarantee returns on loans, and therefore to compete in the world financial market. This leads to a downward (resp. upward) spiral. Under these dynamics the world economy is polarized endogenously into rich and poor countries.
7. Institutions and organizations The fundamental economic problem is scarcity. Since the beginning of life on earth, all organisms have engaged in competition for limited resources. The welfare outcomes of this competition have ranged from efficient allocation to war, genocide and extinction. It is the rules of the game which determine the social welfare consequences. More precisely, it is the long run interaction between the rules of the game and the agents who compete.
364
C. Azariadis and J. Stachurski
Institutions – which make up the rules of the game – were at one time thought to have strong efficiency properties in equilibrium. To a large extent, this is no longer the case [for an introduction to the literature, see, for example, North (1993, 1995); or Hoff (2000)]. Institutions can either reinforce market failure or themselves be the source of inefficiency. Moreover, institutions are path dependent, so that bad equilibria forming from historical accident may be locked in, causing poverty to persist. Among the set of institutions, the state is one of the most important determinants of economic performance; and one of the most common kinds of “government failure” is corruption.71 In Section 7.1 we review why corruption is thought to be not only bad for growth and development, but also self-reinforcing. Section 7.2 then looks at the kinship system, a kind of institution that arises spontaneously in many traditional societies to address such market problems as lack of formal insurance. We consider how these systems may potentially form a local poverty trap, by creating hurdles to adoption of new techniques of production. Although the aggregate outcome is impoverishing, it is shown that the kinship system may nevertheless fail to be dismantled as a result of individual incentives. 7.1. Corruption and rent-seeking Corruption is bad for growth. A number of ways that corruption retards development have been identified in the literature. First, corruption tends to reduce the incentive to invest by decreasing net returns and raising uncertainty. This effect impacts most heavily on increasing returns technologies with large fixed costs. Once costs are sunk, investors are subject to hold-up by corrupt officials, who can extort large sums. Also, governments and officials who have participated in such schemes find it difficult to commit credibly to new infrastructure projects. Second, corruption diverts public expenditure intended for social overhead capital. At the same time, the allocation of such capital is distorted, because officials prefer infrastructure projects where large side payments are feasible. Corruption also hinders the collection of tax revenue, and hence the resource base of the government seeking to provide public infrastructure. Again, a lack of social overhead capital such as transport and communication networks tends to impact more heavily on the modern sector. Third, innovators suffer particularly under a corrupt regime, because of their higher need for such official services as permits, patents and licenses [De Soto (1989), Murphy, Shleifer and Vishny (1993)]. The same is true for foreign investors, who bring in new technology. Lambsdorff (2003) finds that on average a 10% worsening in an index of transparency and corruption he constructs leads to a fall of 0.5 percentage points in the ratio of foreign direct investment to GDP. 71 Following the excellent survey of Bardhan (1997), we define corruption to be “the use of public office
for private gains, where an official (the agent) entrusted with carrying out a task by the public (the principal) engages in some sort of malfeasance for private enrichment which is difficult to monitor for the principal” [Bardhan (1997, p. 1321)].
Ch. 5: Poverty Traps
365
Not only is corruption damaging to growth, but it also tends to breed more corruption. In other words, there are complementarities in corruption and other rent-seeking activities. It is this increasing returns nature of corruption which may serve to lock in poverty. Some equilibria will be associated with high corruption and low income, where many rent-seekers prey on relatively few producers. Others will have the reverse. The decision of one official to seek bribes will increase expected net rewards to bribe taking in several ways. The most obvious of these complementarities is that when many agents are corrupt, the probability of detection and punishment for the marginal official is lowered. A related point is that if corruption is rampant then detection will not entail the same loss of reputation or social stigma as would be the case in an environment where corruption is rare. In other words, corruption is linked to social norms, and is one of the many reasons why they matter for growth.72 Third, greater corruption tends to reduce the search cost for new bribes. Murphy, Shleifer and Vishny (1993) point out yet another source of potential complementarities in rent-seeking. Their idea is that even if returns to predation are decreasing in an absolute sense, they may still be increasing relative to production. This would occur if the returns to productive activities – the alternative when agents make labor supply decisions – fall faster than those to rent-seeking as the number of rent-seekers increases. The general equilibrium effect is that greater rent-seeking decreases the (opportunity) cost of an additional rent-seeker. In their model there is a modern sector, where output by any individual is equal to a, and a subsistence technology with which agents can produce output c < a. Alternatively, agents can prey on workers, obtaining for themselves an amount no more than b per person, but limited by the amount of output available for predation. This in turn depends on the number of people working in the productive sectors. The authors assume, in addition, that only modern sector output can be appropriated by rent-seekers, so returns to subsistence farming are always equal to c. An equilibrium is an allocation of labor across the different occupations such that returns to all are equal, and no individual agent can increase their reward by acting unilaterally. To locate equilibria, we now discuss returns to working in the different sectors as a function of n, which is defined to be the number of rent seekers for each modern sector producer. Returns to employment in the subsistence sector are always given by c. Rentseekers all take a slice b of the pie until their ratio to modern sector producers n satisfies a − bn = c. At this ratio, which we denote n, ¯ the earnings of the modern sector producers fall to that of the subsistence producers, and the rent-seekers must reduce the size of their take (or earn nothing). After n, ¯ the rent-seekers each take (a − c)/n, exactly equalizing returns to modern sector production and subsistence. 72 Transparency International’s 2004 Global Corruption Report cites a statement by the president of the
Government Action Observatory in Burundi that “corruption has spread, openly and publicly, to such an extent that those who practice it have become stronger than those who are fighting against it. This has led to a kind of reversal of values”. (Emphasis added.)
366
C. Azariadis and J. Stachurski
Figure 27.
Let p(n) and r(n) be returns to modern sector production and rent-seeking respectively, so that p(n) = (a − bn)1{n < n} ¯ + c1{n n} ¯ and r(n) = b1{n < n} ¯ + a−c 1{n n}. ¯ These curves are drawn in Figure 27. The figure shows that there are n multiple equilibria whenever the parameters satisfy c < b < a. One is where all work in the modern sector. Then n = 0, and p(n) = p(0) = a > r(n) = b > c. This allocation is an equilibrium, where all agents earn the relatively high revenue available from modern sector production. In addition, because b > c, the payoff functions n → p(n) and n → r(n) intersect above n, ¯ at n2 . This is again an equilibrium, where the payoffs to working in the subsistence sector, the modern sector and the rent-seeking sector are all equal and given by c. Notice that b does not affect income in either of these two equilibria. However, it does affect which one is likely to prevail. If b declines below c, for example, then only the good equilibrium will remain. If it increases above a, then the bad equilibrium will be unique. When there are two equilibria, higher b increases the basin of attraction for the bad equilibrium under myopic Marshallian dynamics.
Ch. 5: Poverty Traps
367
In summary, the model exhibits a general equilibrium complementarity to corruption, which helps illustrate why corruption tends to be self-reinforcing, therefore causing poverty to persist. These kind of stories are important, because in practice corruption and related crimes tend to show a great deal of variation across time and space, often without obvious exogenous characteristics that would cause such variation. There are many other models which exhibit self-reinforcement and path dependence in corruption. One is Tirole (1996), who studies the evolution of individual and group reputation. In his model, past behavior provides information about traits, such as honesty, ability and diligence. However, individual behavior is not perfectly observed. As a result, actions of the group or cohort to which the individual belongs have predictive power when trying to infer the traits of the individual. It follows that outcomes and hence incentives for the individual are affected by the actions of the group. In this case we can imagine the following scenario. Young agents progressively joint an initial cohort of workers, a large number of whom are known to be corrupt. Because the behavior of new agents is imperfectly observed, they inherit the suspicion which already falls on the older workers. As a result, they may have little incentive to act honestly, and drift easily to corruption. This outcome in turn perpetuates the group’s reputation for corrupt action. One can contemplate many more such feedback mechanisms. For example, it is often said that the low wages of petty officials drive them to corruption. But if corruption lowers national output and hence income, then this will reduce the tax base, which in turn decreases the amount of resources with which to pay wages. For further discussion of corruption and poverty traps see Bardhan (1997).73 7.2. Kinship systems All countries and economies are made up of people who at one time were organized in small tribes with their own experiences, customs, taboos and conventions. Over time these tribes were united into cities, states and countries; and the economies within which they operated grew larger and more sophisticated. Some of these economies became vibrant and strong. Others have stagnated. According to North (1993, p. 4): The reason for differing success is straightforward. The complexity of the environment increased as human beings became increasingly interdependent, and more complex institutional structures were necessary to capture the potential gains from trade. Such evolution required that the society develop institutions that will permit anonymous, impersonal exchange across time and space. But to the extent that “local experience” had produced diverse mental models and institutions with respect to the gains from such cooperation, the likelihood of creating the necessary institutions to capture the gains from trade of more complex contracting varied. 73 For other kinds of poverty traps arising through interactions between the state and markets, see, for exam-
ple, Hoff and Stiglitz (2004), or Gradstein (2004).
368
C. Azariadis and J. Stachurski
North and other development thinkers have emphasized that success depends on institutions rewarding efficient, productive activity; and having sufficient flexibility to cope with the structural changes experienced in the transition to modernity. The degree of flexibility and ability to adapt determines to what extent an economy can take advantage of the application of science, of new techniques, and of specialization and the effective division of labor. To illustrate these ideas, in this section we review recent analysis of the “kin” system, an institution found in many traditional societies, usually defined as an informal set of shared rights and obligations between extended family and friends for the purpose of mutual assistance.74 Where markets and state institutions are less developed, the kin system replaces formal insurance and social security by implementing various forms of community risk sharing, and by the provision of other social services [Hoff and Sen (2005)]. The question we ask in the remainder of this section is how, in the process of development, the kin system interacts with the nascent modern sector, and whether or not it may serve to impede the diffusion of new technologies and the exploitation of gains from trade. An interesting example of such analysis is Baker (2004), who interprets Africa’s lack of robust growth as a failure of technology diffusion caused by institutional barriers. She presents a model of a rural African village, and suggests two path dependent mechanisms related to the kin system which may serve to retard growth. Both of them involve community risk sharing, and indicate how technology adoption may have positive network externalities beyond simple social learning. The first mechanism concerns risk sharing among kin members in the form of interest free “loans” with no fixed repayment schedule. Kin members in need can expect to receive these transfers from the better off, who in turn must comply or face various social sanctions (including, in the countries Baker studied, accusations of witchcraft as the source of their good fortune). Beyond the obvious incentive effects on those who might seek to improve their circumstances by using new technology, Baker suggests that a kin member who adopts new techniques may face significant additional uncertainty vis-a-vis income net of transfers if the kin group makes mistakes in estimating his or her true profits. Such a miscalculation may lead to excessive demands for “gifts” or other transfers. As Baker points out, the uncertainty effect of the transfers will be larger for those who adopt new technology, where costs and revenue are harder for the other kin to estimate. For example, the kin may have difficulty in measuring the real costs of new techniques, such as fertilizer or more expensive seed, causing them to overestimate true profits. (New techniques are often associated with higher revenues combined with higher costs.) On the other hand, cost and net profit will be easier to estimate if more kin members have experience of the new techniques. In other words, uncertainty will be mitigated for
74 A related form of local poverty traps is those generated by neighborhood effects. See Durlauf (2004).
Ch. 5: Poverty Traps
369
the marginal adopter if more of his or her fellow kin members adopt the same technology. As a result there are positive network externalities in terms of expected cost. This mechanism generates a coordination problem, whereby a critical mass of co-adopters may be necessary to make the new technology more attractive than the old. This need for coordination may present a barrier to adoption. At the same time, the coordination barrier would not seem to be insurmountable. Perhaps a kin group can negotiate to a better equilibrium when the gains are genuinely large? Baker suggests that in fact this will not be easy, because the risk sharing problem interacts with other path dependent institutions. One of these concerns the nature of old age insurance among self-employed African farmers. Given the lack of state pensions and the difficulty of accumulating assets, support in old age may be contingent on the old providing some form of useful service to the household from which resources are to be acquired. And the most likely candidate for productive service from elderly farmers is the benefit of their experience. The problem here is that the value of this service provided by the old depends on a stagnant technology which does not change from generation to generation. Under new techniques the experience of old farmers may become redundant. If old farmers are able to resist the introduction of new techniques then it will be in their interests to do so. Once again, this is a source of multiple equilibria. The reason is that if the newer technology were already adopted then presumably it would be supported by old farmers, because this is then the methodology in which they have experience. Another interesting study of the kin system has been conducted recently by Hoff and Sen (2005). They analyze the migration of kin members from rural areas to modern sector jobs, and show how network externalities arise in the migration decision. Even if kin members can coordinate on simultaneous migration, Hoff and Sen suggest that the kin group may put up barriers to prevent the loss of their most productive members. It is shown that even when the kin decisions are made by a majority, the barriers can be inefficient in terms of aggregate group welfare. A simplified version of their story runs as follows. Kin members who do migrate may find themselves besieged by their less fortunate brethren. The latter come seeking not only “gifts” of cash transfers, but also help in finding jobs in the modern sector for themselves. Realizing this, employers will find it profitable to restrict employment of kin members. Here we assume these barriers are so high that migration while maintaining kin ties is never optimal. As a result, kin members choose between remaining in the rural sector or migrating while breaking their kin ties. The kin group is thought of as a continuum of members with total mass of one. A fraction α¯ ∈ (0, 1) of the kin receive job offers in the modern sector. The utility of remaining in the rural sector is us (α) = s0 + b(1 − α),
(29)
where here and elsewhere α α¯ is the fraction of the kin who break ties and move. The constant s0 is a stand-alone payoff to rural occupation. The constant b is positive, so that utility of staying is higher when more kin members remain. On the other hand,
370
C. Azariadis and J. Stachurski
Figure 28.
the utility of moving to the modern sector is um (α) = m0 − c(1 − α),
(30)
where m0 is a payoff to working in the modern sector and c is a positive constant. The function α → c(1 − α) is the cost of ending kin membership (measured in the utility equivalent of various social sanctions which we will not describe). It is assumed that the cost of breaking kin ties for the marginal kin member decreases as more members leave the kin group and shift to the modern sector.75
75 Hoff and Sen cite Platteau (2000), who writes that to leave and enter the modern sector, a kin member
“needs the protection afforded by the deviant actions of a sufficient number of other innovators in his locality. Rising economic opportunities alone will usually not suffice to generate dynamic entrepreneurs in the absence of a critical mass of cultural energies harnessed towards countering social resistance . . . ” (Emphasis added.)
Ch. 5: Poverty Traps
371
Consider the interesting case, where um (0) < us (0) and um (α) ¯ > us (α). ¯ A pair of curves for (29) and (30) which fit this pattern are depicted in Figure 28. If no kin members take modern sector jobs then it is not optimal to do so for an individual member. On the other hand, if all those with offers take up jobs, then their utility payoff will be higher than the payoff of those who remain. If, as in the figure, we also have um (α) ¯ > us (0), then it seems plausible that the kin members with job offers will coordinate their way to the equilibrium where all simultaneously move to modern sector jobs. Kin groups are not as diffuse as some other groups of economic actors, and coordination should prove correspondingly less problematic. However, Hoff and Sen show that when kin members are heterogenous, a majority may take steps to forestall coordination by the productive critical mass on movement to the modern sector. Moreover, they may do so even when this choice is inefficient in terms of the kin’s aggregate group payoff. In doing so, the kin group becomes a “dysfunctional institution”, responsible for enforcing an inefficient status quo. Their example works as follows. Consider a two stage game. First, the kin set the exit cost parameter c by majority vote. The two possible values are ca and cb , where ca < cb . Next, job offers are received, and kin members decide whether or not to move. Coordination always takes place in the situation where those with job offers together have a higher payoff in the modern sector. There are now two types of kin members, those with high “ability” and those with low. The first type are of measure γ , and have probability αH of getting a job offer from the modern sector. The second type are of measure 1 − γ , and have probability αL of getting a job offer from the modern sector, where 0 < αL < αH < 1. We assume that γ < 1/2, so high ability types are in the minority. Also, we assume that γ αH + (1 − γ )αL = α. ¯ Ex post, the law of large numbers implies that the fraction of kin members who get job offers will again be α. ¯ Regarding parameters, we assume that uam (α) := m0 − ca (1 − α) satisfies uam (α) ¯ > us (0), but ubm (α) := m0 − cb (1 − α) satisfies ubm (α) ¯ < us (0). The first inequality says that under the low cost regime, the payoff to working in the modern sector is greater than that of staying if all with job offers move. The second inequality says that under the high cost regime the opposite is true. Because of coordination, under ca all of those with job offers will move. The ex ante payoff of the high ability types is ¯ + (1 − αH )us (α), ¯ πHa := αH uam (α) while that of low ability types is ¯ + (1 − αL )us (α). ¯ πLa := αL uam (α) Under cb all remain in the traditional sector, so the payoffs are πLb := us (0) =: πHb . Ex ante aggregate welfare measured as the sum of total payoffs is given under ca by ¯ am (α) ¯ + (1 − α)u ¯ s (α). ¯ Π a := αu Under cb it is Π b := us (0).
372
C. Azariadis and J. Stachurski
Figure 29.
What Hoff and Sen point out is that under some parameters it is possible to have πLa < us (0) = πLb = πHb = Π b < Π a < πHa .
(31)
In this case πLa < us (0) = πLb , and since those with low ability are in the majority they will choose to set c = cb . But then aggregate welfare is reduced, because Π b < Π a . This situation is illustrated in Figure 29.76 Incentives are such that the kinship institution perpetuates a low average income status quo.
76 The parameters are s = 0.8, b = 0.2, m = 2, ca = 1.1, cb = 2.3, α = 0.9, α = 0.1 and γ = 0.45. H L 0 0
Ch. 5: Poverty Traps
373
8. Other mechanisms The poverty trap literature is vast, and even in a survey of this size many models must be neglected. A few of the more egregious omissions are listed in this section. One of the earliest streams of literature on poverty traps is that related to endogenous fertility. A classic contribution is Nelson (1956), who shows how persistent underdevelopment can result from demographics. In his model, any increase in income lowers the death rate, which increases population and lowers capital stock per worker. If the population effect is stronger than diminishing returns then capital per worker cannot rise. See Azariadis (1996, Section 3.4) for other mechanisms and more references. Other kinds of traps that arise in convex economies with complete markets include impatience traps and technology traps. Impatience traps typically involve subsistence levels of consumption, and sensitivity of consumption to income at low levels. See Magill and Nishimura (1984) or Azariadis (1996, Section 3.1). Technology traps are associated with low degrees of substitutability between capital and labor. See Azariadis (1996, Section 3.2). See Dasgupta and Ray (1986) or Dasgupta (2003) for an introduction to the literature on malnutrition and underdevelopment. See also Basu and Van (1998) for a model of child labor with multiple equilibria.
9. Conclusions The poor countries are not rich because they have failed to adopt the modern techniques of production which first emerged in Britain during the Industrial Revolution and then spread to some other nations in Western Europe and elsewhere. As a result, their economies have stagnated. By contrast, the rich countries possess market environments where the same techniques have been continuously refined, upgraded and extended, leading to what are now striking disparities between themselves and the poor. Why would techniques not be adopted even when they are more efficient? Is it not the case that more efficient techniques are more profitable? The main objective of this survey has been to review a large number of studies which show why self-reinforcing traps may prevent the adoption of new technologies. For example, Section 5 showed how increasing returns can generate an incentive structure whereby agents avoid starting modern sector businesses, or invest little in their own training. Section 6 focused on credit market imperfections. Poor individuals lack collateral, which restricts their ability to raise funds. As a result, projects with large fixed costs are beyond the means of the poor, leaving them locked in low return occupations such as subsistence farming. Recently many economists have highlighted the role of institutions in perpetuating poverty. Section 7 looked at why rent-seeking is both bad for growth and yet strongly self-reinforcing. Essentially similar societies may exhibit very different levels of predation simply as a result of historical accident, or some spontaneous coordination of beliefs. In addition, the role of kinship systems was analyzed as representative of the
374
C. Azariadis and J. Stachurski
kinds of social conventions which may potentially harm formation of the modern sector. Together, these mechanisms add up to a very different picture of development than the convex neoclassical benchmark model on which so much of modern growth theory has been based. Growth is not automatic. Small initial differences are magnified and then propagated through time. Poverty coexists with riches, much as it is observed to do in the cross-country income panel. 9.1. Lessons for economic policy There is a real sense in which poverty trap models are optimistic. Poverty is not the result of some simple geographic or cultural determinism. The poor are not condemned to poverty by a set of unfavorable exogenous factors, or even a lack of resources. Temporary policy shocks will have large and permanent effects if one-off interventions can cause the formation of new and better equilibria. In practice, however, engineering the emergence of more efficient equilibria seems problematic for a number of reasons. First, we have seen many examples of how bad equilibria can be stable and self-reinforcing. In this case small policy changes are not enough to escape from their grip. Large changes must be made to the environment that people face, and the structure of their incentives. Such changes may be resisted by the forces that have perpetuated the inefficient equilibrium, such as a corrupt state apparatus fighting to preserve the status quo. Second, coordinating changes in expectations and the status quo is difficult because norms and conventions are highly persistent. While it is possible to change policy and legislation almost instantaneously, it needs to be remembered that informal norms and conventions are often more important in governing behavior than the formal legalistic ones. Informal norms cannot be changed in the manner of interest rates, say, or tariffs. Rather they are determined within the system, and perpetuated by those forces that made them a stable part of the economy’s institutional framework. Third, policies can create new problems as a result of perverse incentives.77 Successful policies will need to be carefully targeted, and operate more on the level of incentives than compulsion. These kinds of policies require a great deal of information. Traps which prevent growth and prosperity cannot be overcome without proper understanding and the careful design of policy.
77 For example, in South Korea the state is generally credited with solving many of the coordination problems
associated with industrialization in that country through their organization and support of large industrial conglomerates, and through active policy-based lending. However, these actions also led to a moral hazard problem, as the industrial groups became highly leveraged with government-backed loans. In the 1970s, investment was increasingly characterized by a costly combination of duplication and poor choices. Losses were massive, and motivated subsequent liberalization.
Ch. 5: Poverty Traps
375
Acknowledgements Support from the Program of Dynamic Economics at UCLA is acknowledged with thanks, as is research assistance from Athanasios Bolmatis, and discussions with David de la Croix, Cleo Fleming, Oded Galor, Karla Hoff, Kirdan Lees and Yasusada Murata. The second author thanks the Center for Operations Research and Econometrics at Université Catholique de Louvain for their hospitality during a period when part of this survey was written. All simulations and estimations use the open source programming language R.
Appendix A Section A.1 gives a general discussion of Markov chains and ergodicity. The proof of Proposition 3.1 is outlined. Section A.2 gives remaining proofs. A.1. Markov chains and ergodicity In the survey we repeatedly made use of a simple framework for treating Markov chains and ergodicity. The following is an elementary review. Our end objective is to sketch the proof of Proposition 3.1, but the review is intended to be more generally applicable. Consider first a discrete time dynamical system evolving in state space S ⊂ Rn . Just as for deterministic systems on S, which are represented by a transition rule associating each point in S with another point in S – the value of the state next period – a Markov chain is represented by a rule associating each point in S with a probability distribution over S. From this conditional distribution (i.e., distribution conditional on the current state x ∈ S) the next period state is drawn. In what follows the conditional distribution will be denoted by Γ (x, dy), where x ∈ S is the current state. Because for Markov chains points in S are mapped into probability distributions rather than into individual points, it seems that the analytical methods used to study the evolution of these processes must be fundamentally different to those used to study deterministic discrete time systems. But this is not the case: Markov chains can always be reduced to deterministic systems. To see this, note that since the state variable xt is now a random variable, it must have some (marginal) distribution on S, which we call ψt . Suppose, as is often the case in economics, that ψt is a density on S, and that the distribution Γ (x, dy) is in fact a density Γ (x, y) dy for every x ∈ S. In that case the marginal distribution for xt+1 is a density ψt+1 , and ψt+1 (y) = S Γ (x, y)ψt (x) dx. This last equality is just a version of the law of total probability: The probability of ending up at y is equal to the probability of going to y via x, weighted by the probability of being at x now, summed over all x ∈ S.
376
C. Azariadis and J. Stachurski
Now define map M : D → D, where D := {ϕ ∈ L1 (S) | ϕ 0 and space of densities on S, by M : D ψ → (Mψ)(·) := Γ (x, ·)ψ(x) dx ∈ D.
ϕ = 1} is the
(A.1)
S
With this definition our law of total probability rule for linking ψt+1 and ψt can be written simply as ψt+1 = Mψt . Since the map M is deterministic, we have succeeded in transforming our stochastic system into a deterministic system to which standard methods of analysis may be applied. The only difficulty is that the state space is now D rather than S. The latter is finite dimensional, while the former clearly is not. The map M is usually called the stochastic operator or Markov operator associated with Γ . There are many good expositions of Markov operators in economics, including Stokey, Lucas and Prescott (1989) and Futia (1982). However those expositions treat the more general case, where Γ (x, dy) does not necessarily have a density representation. Here it does, and it turns out that this extra structure is very useful for treating the models in this survey. We wish to know when the difference equation ψt+1 = Mψt has fixed points, and, more specifically, whether the system is globally stable in the sense that there is a unique fixed point ψ ∗ , and ψt = Mt ψ0 → ψ ∗ as t → ∞ for all ψ0 ∈ D.78 This is just ergodicity in the sense of Definition 3.1. Let · be the L1 norm. Were M a uniform (Banach) contraction on D, which is to say that ∃ λ < 1 with Mψ − Mψ λ ψ − ψ for all ψ, ψ ∈ D, then ergodicity would hold because D is a closed subset of the complete metric space L1 (S). Sadly, for continuous state Markov chains this uniform contraction property rarely holds. However it is often the case that M satisfies a weaker contraction condition: D EFINITION A.1. Let T : X → X, where (X, d) is a metric space. The map T is called a T2 contraction if d(T x, T x ) < d(x, x ) for every x = x in X. T2 contractions maps distinct points strictly closer together. A sufficient condition for M : D → D to satisfy the T2 property is given below. The essential requirement is communication across all regions of the state space. Although T2 contractions do not always have fixed points (examples in R are easy to construct), they do if the state space is compact! In fact if X is a compact set and T : X → X is a T2 contraction then T has unique fixed point x ∗ ∈ X and T t x → x ∗ as t → ∞ for all x ∈ X. This is just what we require for ergodicity when M is thought of as a map on D. Now D is not itself a compact set in the L1 norm topology, but it may be the case that every orbit (Mt ψ0 )t0 of M is compact when taken with its closure. (From now on, call
78 Here Mt is t compositions of M with itself, and ψ is the marginal distribution of x , so iterating the 0 0 difference equation backwards gives ψt = Mt ψ0 .
Ch. 5: Poverty Traps
377
a set with compact closure precompact.) Such a property is called Lagrange stability.79 And it turns out that Lagrange stability can substitute for compactness of the state space D: If M is (a) a T2 contraction, and (b) Lagrange stable, then the associated Markov chain is ergodic.80 How to establish Lagrange stability? To check precompactness of orbits it seems we must look at characterizations of compactness in L1 (there is a famous one due to Kolmogorov), but Lasota (1994, Theorem 4.1) has proved that one need only check weak precompactness.81 In fact it is sufficient to check weak precompactness of orbits starting from ψ ∈ D0 , where D0 is a (norm) dense subset of D. Weak compactness is much easier to work with than norm compactness. Several well-known conditions are available. Using one such condition due to Dunford and Pettis, Mirman, Reffett and Stachurski (2005) show that Lasota’s criterion for Lagrange stability is satisfied when (i) there exists a continuous “norm-like” function V : S → R and constants α, β ∈ [0, ∞), α < 1, such that Γ (x, y)V (y) dy αV (x) + β, ∀x ∈ S; (A.2) and (ii) there exists a continuous function h : S → R such that supx∈S Γ (x, y) h(y) for all y ∈ S. By V being norm-like is meant that V is nonnegative, and that the sets {x ∈ S: V (x) a} are precompact for all a. (For example, when S = Rn it is easy to convince yourself that x → x is norm-like. Note that when S is a proper subset of Rn precompactness of sublevel sets refers to the relative Euclidean topology on S.) Condition (i) is a standard drift condition, which pushes probability mass towards the center of the state space. This implies that orbits of the Markov process will be “tight”. Tightness is a component of Dunford and Pettis’ criterion for weak precompactness. Condition (ii) is just a technical condition which combines with (i) to fill out the requirements of the Dunford–Pettis criterion. In the case of Proposition 3.1, we can take S = (0, ∞), where 0 ∈ / S so that any stationary distribution we find is automatically nontrivial. One can then show that V (x) = |ln x| is norm-like on S, and a little bit of algebra shows that condition (i) holds for Γ given in (10). Also, one can show that (ii) holds when h(y) := 1/y.82 This takes care of Lagrange stability. Regarding T2 contractiveness, one can show that M is a T2 contraction whenever the set supp Mψ ∩ supp Mψ has positive measure for all ψ, ψ ∈ D, where supp f := {x ∈ S | f (x) = 0}. This basically says that 79 That is, a self-mapping T on topological space X is called Lagrange stable if the set {T t x | t 0} is
precompact for every x ∈ X.
80 The proof that Lagrange stability is sufficient is not hard. See Stachurski (2002, Theorem 5.2). 81 Here is where the density structure is crucial. The operator M inherits nice properties from the fact that
Γ (x, dy) has a density representation. Also, we can work in L1 rather than a space of measures. The former has a nice norm-dual space in L∞ – helpful when dealing with weak precompactness. 82 For more details see Stachurski (2004).
378
C. Azariadis and J. Stachurski
probability mass is mixed across the state space – all areas of S communicate. In the case of (10) it is easy to show that supp Mψ = (0, ∞) = S for every ψ ∈ D. This is clearly sufficient for the condition. A.2. Remaining proofs The proof of Proposition 5.1 in Section 5.2 is now given. The first point is that the banks b = 1, . . . , B − 1 are equal-cost Bertrand competitors, and as a result always offer the interest rate r to all firms in equilibrium. The main issue is the optimal strategy of the last bank B. So consider the following strategy σB∗ for B, which is illustrated with the help of Figure 30. To firm n the bank offers in∗ defined by in∗ = f [(n − 1)/N] − 1 if n αC N. To the remaining firms B offers the interest rate r. (Without loss of generality, we suppose that the index of firms from 1 to N and the ranking of the offers made by
Figure 30.
Ch. 5: Poverty Traps
379
B always coincide.) Let σ ∗ ∈ Σ be the strategy where B offers σB∗ and all other banks offer r. For the strategy σ ∗ we have Ω(σ ∗ ) = {1}. The reason is that for α = n/N αC , firms j = 1, . . . , n + 1 all satisfy ∗ π n/N, mj (σ ∗ ) π n/N, ij∗ π n/N, in+1 = 0. In which case α ∈ / Ω(σ ∗ ) by (20). Also, for α ∈ (αC , 1) we have π(α, mn (σ ∗ )) π(α, r) 0 for all n ∈ {1, . . . , N }, so again α ∈ / Ω(σ ∗ ). For the same reason, 1 ∈ ∗ ∗ Ω(σ ), because π(1, mn (σ )) π(1, r) 0. It follows that under this strategy α pes (σ ∗ ) = 1. By (19) all firms enter. The profits of bank B are given by the sum of the regions P , Q and R, minus the region O, in Figure 30. Here Q and α¯ are chosen so that P + Q − O = 0. Thus, α¯ is the break-even point for the bank, where it recoups all losses made by offering cheap loans to firms in the “critical mass” region [0, αC ]. If /N α¯ and hence R 0, the bank B makes positive profits. It is not too hard to see that σ ∗ is indeed the optimal strategy in Σ for the banks. The banks b = 1, . . . , B − 1 always offer r. For B, strategy σB∗ is optimal for the following reasons. First, if B offers interest rates to n ∈ {1, . . . , N } which are all less than or equal to those in σB∗ , then all firms will enter as above, but B will make lower profits ∗∗ } where i ∗∗ > i ∗ for by (17). So suppose that B offers a schedule of rates {i1∗∗ , . . . , iN n n at least one n, and let k be the first such n. It is not difficult to see that the chain of logic whereby all firms enter now unravels: It must be that k/N αC , because to other firms B offers the rate r, which cannot be exceeded due to B’s competitors. One can now check that (k − 1)/N ∈ Ω(σ ∗∗ ), and in fact (k − 1)/N = min Ω(σ ∗∗ ). As a result, α pes (σ ∗∗ ) = (k − 1)/N, and precisely k − 1 firms enter. Clearly the profits of B are lower for σ ∗∗ than for σ ∗ .
References Acemoglu, D. (1997). “Training and innovation in an imperfect labor market”. Review of Economic Studies 64 (3), 445–464. Acemoglu, D., Johnson, S., Robinson, J. (2005). “Institutions as the fundamental cause of long-run growth”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1A. Elsevier, Amsterdam. Chapter 6 in this volume. Acemoglu, D., Zilibotti, F. (1997). “Was Prometheus unbound by chance? Risk, diversification and growth”. Journal of Political Economy 105 (4), 709–751. Adsera, A., Ray, D. (1997). “History and coordination failure”. Journal of Economic Growth 3, 267–276. Aghion, P., Bolton, P. (1997). “A theory of trickle-down growth and development”. Review of Economic Studies 64, 151–172. Amir, R., Mirman, L.J., Perkins, W.R. (1991). “One-sector nonclassical optimal growth: optimality conditions and comparative dynamics”. International Economic Review 32 (3), 625–644. Arthur, W.B. (1994). Increasing Returns and Path Dependence in the Economy. University of Michigan Press, Ann Arbor. Azariadis, C. (1996). “The economics of poverty traps”. Journal of Economic Growth 1, 449–486.
380
C. Azariadis and J. Stachurski
Azariadis, C. (2005). “The theory of poverty traps: what have we learned?”. In: Bowles, S., Durlauf, S., Hoff, K. (Eds.), Poverty Traps. Princeton University Press, Princeton. Azariadis, C., Drazen, A. (1990). “Threshold externalities in economic development”. Quarterly Journal of Economics 105, 501–526. Azariadis, C., Stachurski, J. (2004). “A forward projection of the cross-country income distribution”. Mimeo, Université Catholique de Louvain. Baker, E. (2004). “Institutional barriers to technology adoption in rural Africa”. Mimeo, Stanford University. Bandiera, O., Rasul, I. (2003). “Complementarities, social networks and technology adoption in northern Mozambique”. Mimeo. Banerjee, A. (2003). “The two poverties”. Mimeo, Massachusetts Institute of Technology. Banerjee, A., Newman, A. (1993). “Occupational choice and the process of development”. Journal of Political Economy 101, 274–298. Bardhan, P. (1997). “Corruption and development: a review of the issues”. Journal of Economic Literature 35, 1320–1346. Barrett, C.B., Bezuneh, M., Aboud, A. (2001). “Income diversification, poverty traps and policy shocks in Côte d’Ivoire and Kenya”. Food Policy 26 (4), 367–384. Barrett, C.B., Swallow, B.M. (2003). “Fractal poverty traps”. Strategies and Analysis for Growth and Access Working Paper, Cornell and Clark Atlanta Universities. Basu, K., Van, P.H. (1998). “The economics of child labor”. American Economic Review 88 (3), 412–427. Beath, J., Katsoulacos, Y., Ulph, D. (1995). “Game-theoretic approaches to the modeling of technological change”. In: Stoneman, P. (Ed.), Handbook of the Economics of Innovation and Technological Change. Blackwell. Bianchi, M. (1997). “Testing for convergence: evidence from nonparametric multimodality tests”. Journal of Applied Econometrics 12, 393–409. Bloom, D.E., Canning, D., Sevilla, J. (2003). “Geography and poverty traps”. Journal of Economic Growth 8, 355–378. Bowles, S., Durlauf, S.N., Hoff, K. (Eds.) (2005). Poverty Traps. Princeton University Press, Princeton. Brock, W.A., Mirman, L. (1972). “Optimal economic growth and uncertainty: the discounted case”. Journal of Economic Theory 4, 479–513. Caballero, R., Lyons, R.K. (2002). “The case for external economies”. In: Cukierman, A., et al. (Eds.), Political Economy, Growth and Business Cycles. MIT Press, Cambridge, MA. Cass, D. (1965). “Optimum growth in an aggregative model of capital accumulation”. Review of Economic Studies 32, 233–240. Checchi, D., García-Peñalosa, C. (2004). “Risk and the distribution of human capital”. Economics Letters 82 (1), 53–61. Chen, S., Ravallion, M. (2001). “How did the world’s poor fare in the 1990s?”. Review of Income and Wealth 47 (3), 238–300. Conley, T.G., Udry, C.R. (2003). “Learning about a new technology: pinapple in Ghana.” Mimeo. Da Rin, M., Hellmann, T. (2002). “Banks as catalysts for industrialization”. Journal of Financial Intermediation 11, 366–397. Dasgupta, P. (2003). “World poverty: causes and pathways”. Mimeo, University of Cambridge. Dasgupta, P., Ray, D. (1986). “Inequality as a determinant of malnutrition and unemployment”. The Economic Journal 96, 1011–1034. David, P.A. (1994). “Why are institutions the ‘carriers of history’?: Path dependence and the evolution of conventions, organizations and institutions”. Structural Change and Economic Dynamics 5 (2), 205–220. Dechert, W.D., Nishimura, K. (1983). “A complete characterization of optimal growth paths in an aggregated model with non-concave production function”. Journal of Economic Theory 31, 332–354. Den Haan, W.J. (1995). “Convergence of stochastic growth models: the importance of understanding why income levels differ”. Journal of Monetary Economics 35, 65–82. Dercon, S. (1998). “Wealth, risk and activity choice: cattle in Western Tanzania”. Journal of Development Economics 55, 1–42.
Ch. 5: Poverty Traps
381
Dercon, S. (2003). “Risk and poverty: a selective review of the issues”. Mimeo, Oxford University. Desdoigts, A. (1999). “Patters of economic development and the formation of clubs”. Journal of Economic Growth 4 (3), 305–330. De Soto, H. (1989). The Other Path: The Invisible Revolution in the Third World. Harper and Row, New York. Dimaria, C.H., Le Van, C. (2002). “Optimal growth, debt, corruption and R&D”. Macroeconomic Dynamics 6, 597–613. Durlauf, S.N. (1993). “Nonergodic economic growth”. Review of Economic Studies 60, 349–366. Durlauf, S.N. (2004). “Neighborhood effects”. In: Henderson, J.V., Thisse, J.F. (Eds.), Handbook of Regional and Urban Economics, vol. 4. Elsevier, Amsterdam. Durlauf, S.N., Johnson, P.A. (1995). “Multiple regimes and cross-country growth behavior”. Journal of Applied Econometrics 10 (4), 365–384. Easterly, W. (2001). The Elusive Quest for Growth: Economists’ Adventures and Misadventures in the Tropics. MIT Press, Cambridge, MA. Easterly, W., Kremer, M., Pritchett, L., Summers, L.H. (1993). “Good policy or good luck: country growth performance and temporary shocks”. Journal of Monetary Economics 32, 459–484. Easterly, W., Levine, R. (2000). “It’s not factor accumulation: stylized facts and growth models”. World Bank Economic Review 15 (2), 177–219. Engerman, S., Sokoloff, K.L. (2005). “The persistence of poverty in the Americas: the role of institutions”. In: Bowles, S., Durlauf, S., Hoff, K. (Eds.), Poverty Traps. Princeton University Press, Princeton. Feyrer, J. (2003). “Convergence by parts”. Mimeo, Dartmouth College. Freeman, S. (1996). “Equilibrium income inequality among identical agents”. Journal of Political Economy 104 (5), 1047–1064. Futia, C.A. (1982). “Invariant distributions and the limiting behavior of Markovian economic models”. Econometrica 50, 377–408. Galor, O., Zeira, J. (1993). “Income distribution and macroeconomics”. Review of Economic Studies 60, 35–52. Glynn, P.W., Henderson, S.G. (2001). “Computing densities for Markov chains via simulation”. Mathematics of Operations Research 26, 375–400. Gradstein, M. (2004). “Governance and growth”. Journal of Development Economics 73, 505–518. Graham, B.S., Temple, J. (2004). “Rich nations poor nations: how much can multiple equilibria explain?”. Mimeo (revised version of CEPR Discussion Paper 3046). Greif, A., Milgrom, P., Weingast, B.R. (1994). “Coordination, commitment and enforcement: the case of merchant guilds”. Journal of Political Economy 102 (4), 745–776. Hall, R.E., Jones, C.I. (1999). “Why do some countries produce so much more output per worker than others?”. Quarterly Journal of Economics 114 (1), 84–116. Heston, A., Summers, R., Aten, B. (2002). “Penn World Table version 6.1”. Center for International Comparisons, University of Pennsylvania. Hoff, K. (2000). “Beyond Rosenstein–Rodan: the modern theory of coordination problems in development”. In: Proceedings of the World Bank Annual Conference on Development Economics 2000. World Bank, Washington, DC. Hoff, K., Pandey, P. (2004). “Belief systems and durable inequalities: an experimental investigation of Indian caste”. World Bank Policy Research Working Paper 3351. Hoff, K., Sen, A. (2005). “The kin system as a poverty trap”. In: Bowles, S., Durlauf, S., Hoff, K. (Eds.), Poverty Traps. Princeton University Press, Princeton. Hoff, K., Stiglitz, J. (2004). “After the Big Bang? Obstacles to the emergence of the rule of law in postcommunist societies”. American Economic Review 94 (3), 753–763. Jalan, J., Ravallion, M. (2002). “Geographic poverty traps? A micro model of consumption growth in rural China”. Journal of Applied Econometrics 17, 329–346. Johnson, P.A. (2004). “A continuous state space approach to convergence by parts”. Mimeo, Vassar College. Kehoe, T., Levine, D. (1993). “Debt constrained asset markets”. Review of Economic Studies 60, 865–888.
382
C. Azariadis and J. Stachurski
King, R.G., Rebelo, S.T. (1993). “Transitional dynamics and economic growth in the neoclassical model”. American Economic Review 83 (4), 908–931. Kiyotaki, N., Moore, J.H. (1997). “Credit cycles”. Journal of Political Economy 105 (2), 211–248. Koopmans, T. (1965). “On the concept of optimal economic growth”. Pontificae Academiae Scientiarum Scripta Varia 28, 225–300. Kremer, M. (1993). “The O-ring theory of economic development”. Quarterly Journal of Economics 108 (3), 551–575. Krugman, P. (1991). “History versus expectations”. The Quarterly Journal of Economics 106 (2), 651–667. Lambsdorff, J.G. (2003). “How corruption affects persistent capital flows”. Economics of Governance 4 (3), 229–243. Lasota, A. (1994). “Invariant principle for discrete time dynamical systems”. Universitatis Iagellonicae Acta Mathematica 31, 111–127. Limao, N., Venables, A.J. (2001). “Infrastructure, geographical disadvantage, transport costs and trade”. World Bank Economic Review 15, 451–479. Ljungqvist, L. (1993). “Economic underdevelopment: the case of the missing market for human capital”. Journal of Development Economics 40, 219–239. Loury, G.C. (1981). “Intergenerational transfers and the distribution of earnings”. Econometrica 49, 843–867. Lucas, R.E. Jr (1986). “Adaptive behavior and economic theory”. Journal of Business 59 (4), 401–426. Lucas, R.E. (1990). “Why doesn’t capital flow from rich to poor countries?”. American Economic Review 80 (2), 92–96. Maddison, A. (1995). Monitoring the World Economy. OECD Development Center, Paris. Magill, M., Nishimura, K. (1984). “Impatience and accumulation”. Journal of Mathematical Analysis and Applications 98, 270–281. Majumdar, M., Mitra, T., Nyarko, Y. (1989). “Dynamic optimization under uncertainty: non-convex feasible set”. In: Feiwel, G.R. (Ed.), Joan Robinson and Modern Economic Theory. MacMillan, New York. Matsuyama, K. (1991). “Increasing returns, industrialization and indeterminacy of equilibrium”. Quarterly Journal of Economics 106, 617–650. Matsuyama, K. (1995). “Complementarities and cumulative processes in models of monopolistic competition”. Journal of Economic Literature 33, 701–729. Matsuyama, K. (1997). “Complementarity, instability and multiplicity”. Japanese Economic Review 48 (3), 240–266. Matsuyama, K. (2000). “Endogenous inequality”. Review of Economic Studies 67, 743–759. Matsuyama, K. (2004). “Financial market globalization, symmetry-breaking, and endogenous inequality of nations”. Econometrica 72, 853–884. Matsuyama, K., Ciccone, A. (1996). “Start-up costs and pecuniary externalities as barriers to economic development”. Journal of Development Economics 49, 33–59. Mirman, L.J., Morand, O.F., Reffett, K. (2004). “A qualitative approach to Markovian equilibrium in infinite horizon economies with capital”. Mimeo, Arizona State University. Mirman, L.J., Reffett, K., Stachurski, J. (2005). “Some stability results for Markovian economic semigroups”. International Journal of Economic Theory 1 (1). Mokyr, J. (2002). The Gifts of Athena: Historical Origins of the Knowledge Economy. Princeton University Press, Princeton, NJ. Mookherjee, D., Ray, D. (2001). Readings in the Theory of Economic Development. Blackwell, New York. Mookherjee, D., Ray, D. (2003). “Persistent inequality”. Review of Economic Studies 70, 369–393. Morduch, J. (1990). “Risk, production and saving: Theory and evidence from Indian households”. Mimeo, Harvard University. Murphy, K.M., Shleifer, A., Vishny, R.W. (1989). “Industrialization and the big push”. Journal of Political Economy 97, 1003–1026. Murphy, K.M., Shleifer, A., Vishny, R.W. (1993). “Why is rent-seeking so costly to growth?”. American Economic Review 83 (2), 409–414. Nelson, R.R. (1956). “A theory of the low level equilibrium trap”. American Economic Review 46, 894–908.
Ch. 5: Poverty Traps
383
Nishimura, K., Stachurski, J. (2004). “Stability of stochastic optimal growth models: a new approach”. Journal of Economic Theory, in press. North, D.C. (1993). “The new institutional economics and development”. WUSTL Economics Working Paper Archive. North, D.C. (1995). “Some fundamental puzzles in economic history/development”. WUSTL Economics Working Paper Archive. Nurkse, R. (1953). Problems of Capital-Formation in Underdeveloped Countries, 1962 ed. Oxford University Press, New York. Ottaviano, G., Thisse, J.F. (2004). “Agglomeration and economic geography”. In: Henderson, J.V., Thisse, J.F. (Eds.), Handbook of Regional and Urban Economics, vol. 4. Elsevier, Amsterdam. Parente, S.L., Prescott, E.C. (2005). “A unified theory of the evolution of international income levels”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1B. Elsevier, Amsterdam. Piketty, T. (1997). “The dynamics of the wealth distribution and the interest rate with credit rationing”. Review of Economic Studies 64, 173–189. Platteau, J.-P. (2000). Institutions, Social Norms and Economic Development. Harwood Publishers, Amsterdam. Prescott, E.C. (1998). “Needed: a theory of total factor productivity”. International Economic Review 39, 529–549. Pritchett, L. (1997). “Divergence, big time”. Journal of Economic Perspectives 11 (3), 3–17. Quah, D.T. (1993). “Empirical cross-section dynamics in economic growth”. European Economic Review 37, 426–434. Quah, D. (1996). “Convergence empirics across economies with (some) capital mobility”. Journal of Economic Growth 1, 95–124. Radlet, S. (2004). “Aid effectiveness and the Millennium development goals”. Manuscript prepared for the Millennium Project Task Force, United Nations Development Group. Ray, D. (1990). “Income distribution and macroeconomic behavior”. Mimeo, New York University. Ray, D. (2003). “Aspirations, poverty and economic change”. Mimeo, New York University. Ray, D., Streufert, P. (1993). “Dynamic equilibria with unemployment due to undernourishment”. Economic Theory 3, 61–85. Redding, S., Venables, A.J. (2004). “Economic geography and international inequality”. Journal of International Economics 62, 53–82. Rodríguez-Clare, A. (1996). “The division of labor and economic development”. Journal of Development Economics 49, 3–32. Rodrik, D. (1996). “Coordination failures and government policy: A model with applications to East Asia and Eastern Europe”. Journal of International Economics 40, 1–22. Romer, P.M. (1986). “Increasing returns and long-run growth”. Journal of Political Economy 94 (5), 1002– 1037. Romer, P.M. (1990). “Are nonconvexities important for understanding growth?”. American Economic Review 80 (2), 97–103. Rosenstein-Rodan, P. (1943). “The problem of industrialization of Eastern and South-Eastern Europe”. Economic Journal 53, 202–211. Rostow, W.W. (1975). How it All Began: Origins of the Modern Economy. McGraw-Hill, New York. Rostow, W.W. (1990). The Stages of Economic Growth: A Noncommunist Manifesto, third ed. Cambridge University Press, Cambridge. Sachs, J.D., McArthur, J.W., Schmidt-Traub, G., Kruk, M., Bahadur, C., Faye, M., McCord, G. “Ending Africa’s poverty trap”. Mimeo. Simon, H.A. (1986). “Rationality in psychology and economics”. Journal of Business 59 (4), 209–224. Skiba, A.K. (1978). “Optimal growth with a convex-concave production function”. Econometrica 46, 527– 539. Solow, R.M. (1956). “A contribution to the theory of economic growth”. Quarterly Journal of Economics 70, 65–94.
384
C. Azariadis and J. Stachurski
Stachurski, J. (2002). “Stochastic optimal growth with unbounded shock”. Journal of Economic Theory 106, 45–60. Stachurski, J. (2004). “Stochastic economic dynamics”. Mimeo, The University of Melbourne. Starrett, D. (1978). “Market allocations of location choice in a model with free mobility”. Journal of Economic Theory 17, 21–37. Stokey, N.L., Lucas, R.E., Prescott, E.C. (1989). Recursive Methods in Economic Dynamics. Harvard University Press, Cambridge, MA. Sugden, R. (1989). “Spontaneous order”. Journal of Economic Perspectives 3 (4), 85–97. Tirole, J. (1996). “A theory of collective reputations (with applications to persistence of corruption and to firm quality)”. Review of Economic Studies 63, 1–22. Tsiddon, D. (1992). “A moral hazard trap to growth”. International Economic Review 33 (2), 299–321. Van Huyck, J.B., Cook, J.P., Battalio, R.C. (1997). “Adaptive behavior and coordination failure”. Journal of Economic Behavior and Organization 32 (4), 483–503. Young, A.A. (1928). “Increasing returns and economic progress”. Economic Journal 28, 527–542. Zilibotti, F. (1995). “A Rostovian model of endogenous growth and underdevelopment traps”. European Economic Review 39, 1569–1602.
Chapter 6
INSTITUTIONS AS A FUNDAMENTAL CAUSE OF LONG-RUN GROWTH DARON ACEMOGLU Department of Economics, MIT, 50 Memorial Drive E52-380b, Cambridge, MA 02142 e-mail:
[email protected] SIMON JOHNSON Sloan School of Management, MIT, 50 Memorial Drive, Cambridge, MA 02142 e-mail:
[email protected] JAMES A. ROBINSON Department of Government, WCFIA, Harvard University, 1033 Massachusetts Avenue, Cambridge, MA 02138 e-mail:
[email protected]
Contents Abstract Keywords 1. Introduction 1.1. The question 1.2. The argument 1.3. Outline
2. Fundamental causes of income differences 2.1. Three fundamental causes 2.1.1. Economic institutions 2.1.2. Geography 2.1.3. Culture
3. Institutions matter 3.1. The Korean experiment 3.2. The colonial experiment
4. The Reversal of Fortune 4.1. The reversal among the former colonies 4.2. Timing of the reversal 4.3. Interpreting the reversal 4.4. Economic institutions and the reversal
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01006-3
386 387 388 388 389 396 396 397 397 399 400 402 404 407 407 408 412 412 414
386
D. Acemoglu et al. 4.5. Understanding the colonial experience 4.6. Settlements, mortality and development
5. Why do institutions differ? 5.1. 5.2. 5.3. 5.4.
The efficient institutions view – the Political Coase Theorem The ideology view The incidental institutions view The social conflict view
6. Sources of inefficiencies 6.1. 6.2. 6.3. 6.4. 6.5. 6.6. 6.7.
Hold-up Political losers Economic losers The inseparability of efficiency and distribution Comparative statics The colonial experience in light of the comparative statics Reassessment of the social conflict view
7. The social conflict view in action 7.1. 7.2. 7.3. 7.4.
Labor markets Financial markets Regulation of prices Political power and economic institutions
8. A theory of institutions 8.1. Sources of political power 8.2. Political power and political institutions 8.3. A theory of political institutions
9. The theory in action 9.1. 9.2. 9.3. 9.4.
Rise of constitutional monarchy and economic growth in early modern Europe Summary Rise of electoral democracy in Britain Summary
10. Future avenues Acknowledgements References
416 417 421 422 424 425 427 428 430 432 434 436 437 438 439 439 440 441 443 445 448 448 449 451 452 452 457 458 462 463 464 464
Abstract This paper develops the empirical and theoretical case that differences in economic institutions are the fundamental cause of differences in economic development. We first document the empirical importance of institutions by focusing on two “quasi-natural experiments” in history, the division of Korea into two parts with very different economic institutions and the colonization of much of the world by European powers starting in the fifteenth century. We then develop the basic outline of a framework for thinking about why economic institutions differ across countries. Economic institutions determine the incentives of and the constraints on economic actors, and shape economic
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
387
outcomes. As such, they are social decisions, chosen for their consequences. Because different groups and individuals typically benefit from different economic institutions, there is generally a conflict over these social choices, ultimately resolved in favor of groups with greater political power. The distribution of political power in society is in turn determined by political institutions and the distribution of resources. Political institutions allocate de jure political power, while groups with greater economic might typically possess greater de facto political power. We therefore view the appropriate theoretical framework as a dynamic one with political institutions and the distribution of resources as the state variables. These variables themselves change over time because prevailing economic institutions affect the distribution of resources, and because groups with de facto political power today strive to change political institutions in order to increase their de jure political power in the future. Economic institutions encouraging economic growth emerge when political institutions allocate power to groups with interests in broad-based property rights enforcement, when they create effective constraints on power-holders, and when there are relatively few rents to be captured by power-holders. We illustrate the assumptions, the workings and the implications of this framework using a number of historical examples.
Keywords institutions, growth, development, political power, rents, conflict, property rights, efficiency, distributions JEL classification: D7, H1, O10, O40
388
D. Acemoglu et al.
1. Introduction 1.1. The question The most trite yet crucial question in the field of economic growth and development is: Why are some countries much poorer than others? Traditional neoclassical growth models, following Solow (1956), Cass (1965) and Koopmans (1965), explain differences in income per capita in terms of different paths of factor accumulation. In these models, cross-country differences in factor accumulation are due either to differences in saving rates (Solow), preferences (Cass–Koopmans), or other exogenous parameters, such as total factor productivity growth. In these models there are institutions, for example agents have well defined property rights and exchange goods and services in markets, but differences in income and growth are not explained by variation in institutions. The first wave of the more recent incarnations of growth theory, following Romer (1986) and Lucas (1988) differed in the sense that they emphasized that externalities from physical and human capital accumulation could induce sustained steady-state growth. However, they also stayed squarely within the neoclassical tradition of explaining differences in growth rates in terms of preferences and endowments. The second wave of models, particularly Romer (1990), Grossman and Helpman (1991) and Aghion and Howitt (1992), endogenized steady-state growth and technical progress, but their explanation for income differences is similar to that of the older theories. For instance, in the model of Romer (1990), a country may be more prosperous than another if it allocates more resources to innovation, but what determines this is essentially preferences and properties of the technology for creating ‘ideas’.1 Though this theoretical tradition is still vibrant in economics and has provided many insights about the mechanics of economic growth, it has for a long time seemed unable to provide a fundamental explanation for economic growth. As North and Thomas (1973, p. 2) put it: “the factors we have listed (innovation, economies of scale, education, capital accumulation, etc.) are not causes of growth; they are growth” (italics in original). Factor accumulation and innovation are only proximate causes of growth. In North and Thomas’s view, the fundamental explanation of comparative growth is differences in institutions. What are institutions exactly? North (1990, p. 3) offers the following definition: “Institutions are the rules of the game in a society or, more formally, are the humanly devised constraints that shape human interaction”. He goes on to emphasize the key implications of institutions since, “In consequence they structure incentives in human exchange, whether political, social, or economic”.
1 Although, as we discuss later, some recent contributions to growth theory emphasize the importance of economic policies, such as taxes, subsidies to research, barriers to technology adoption and human capital policy, they typically do not present an explanation for why there are differences in these policies across countries.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
389
Of primary importance to economic outcomes are the economic institutions in society such as the structure of property rights and the presence and perfection of markets. Economic institutions are important because they influence the structure of economic incentives in society. Without property rights, individuals will not have the incentive to invest in physical or human capital or adopt more efficient technologies. Economic institutions are also important because they help to allocate resources to their most efficient uses, they determine who gets profits, revenues and residual rights of control. When markets are missing or ignored (as they were in the Soviet Union, for example), gains from trade go unexploited and resources are misallocated. Societies with economic institutions that facilitate and encourage factor accumulation, innovation and the efficient allocation of resources will prosper. Central to this chapter and to much of political economy research on institutions is that economic institutions, and institutions more broadly, are endogenous; they are, at least in part, determined by society, or a segment of it. Consequently, the question of why some societies are much poorer than others is closely related to the question of why some societies have much “worse economic institutions” than others. Even though many scholars including John Locke, Adam Smith, John Stuart Mill, Arthur Lewis, Douglass North and Robert Thomas, and recently many papers in the literature on economic growth and development, have emphasized the importance of economic institutions, we are far from a useful framework for thinking about how economic institutions are determined and why they vary across countries. In other words, while we have good reason to believe that economic institutions matter for economic growth, we lack the crucial comparative static results which will allow us to explain why equilibrium economic institutions differ (and perhaps this is part of the reason why much of the economics literature has focused on the proximate causes of economic growth, largely neglecting fundamental institutional causes). This chapter has three aims. First, we selectively review the evidence that differences in economic institutions are a fundamental cause of cross-country differences in prosperity. Second, we outline a framework for thinking about why economic institutions vary across countries. We emphasize the potential comparative static results of this framework and also illustrate the key mechanisms through a series of historical examples and case studies. Finally, we highlight a large number of areas where we believe future theoretical and empirical work would be very fruitful. 1.2. The argument The basic argument of this chapter can be summarized as follows: 1. Economic institutions matter for economic growth because they shape the incentives of key economic actors in society, in particular, they influence investments in physical and human capital and technology, and the organization of production. Although cultural and geographical factors may also matter for economic performance, differences in economic institutions are the major source of cross-country differences in economic growth and prosperity. Economic institutions not only determine the ag-
390
D. Acemoglu et al.
gregate economic growth potential of the economy, but also an array of economic outcomes, including the distribution of resources in the future (i.e., the distribution of wealth, of physical capital or human capital). In other words, they influence not only the size of the aggregate pie, but how this pie is divided among different groups and individuals in society. We summarize these ideas schematically as (where the subscript t refers to current period and t + 1 to the future): economic performancet . economic institutionst ⇒ distribution of resourcest+1 2. Economic institutions are endogenous. They are determined as collective choices of the society, in large part for their economic consequences. However, there is no guarantee that all individuals and groups will prefer the same set of economic institutions because, as noted above, different economic institutions lead to different distributions of resources. Consequently, there will typically be a conflict of interest among various groups and individuals over the choice of economic institutions. So how are equilibrium economic institutions determined? If there are, for example, two groups with opposing preferences over the set of economic institutions, which group’s preferences will prevail? The answer depends on the political power of the two groups. Although the efficiency of one set of economic institutions compared with another may play a role in this choice, political power will be the ultimate arbiter. Whichever group has more political power is likely to secure the set of economic institutions that it prefers. This leads to the second building block of our framework: political powert ⇒ economic institutionst . 3. Implicit in the notion that political power determines economic institutions is the idea that there are conflicting interests over the distribution of resources and therefore indirectly over the set of economic institutions. But why do the groups with conflicting interests not agree on the set of economic institutions that maximize aggregate growth (the size of the aggregate pie) and then use their political power simply to determine the distribution of the gains? Why does the exercise of political power lead to economic inefficiencies and even poverty? We will explain that this is because there are commitment problems inherent in the use of political power. Individuals who have political power cannot commit not to use it in their best interests, and this commitment problem creates an inseparability between efficiency and distribution because credible compensating transfers and side-payments cannot be made to offset the distributional consequences of any particular set of economic institutions. 4. The distribution of political power in society is also endogenous, however. In our framework, it is useful to distinguish between two components of political power, which we refer to as de jure (institutional) and de facto political power. Here de jure political power refers to power that originates from the political institutions in society. Political institutions, similarly to economic institutions, determine the constraints on and the incentives of the key actors, but this time in the political sphere. Examples of political institutions include the form of government, for example, democracy vs. dictatorship or
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
391
autocracy, and the extent of constraints on politicians and political elites. For example, in a monarchy, political institutions allocate all de jure political power to the monarch, and place few constraints on its exercise. A constitutional monarchy, in contrast, corresponds to a set of political institutions that reallocates some of the political power of the monarch to a parliament, thus effectively constraining the political power of the monarch. This discussion therefore implies that: political institutionstt ⇒ de jure political powert . 5. There is more to political power than political institutions, however. A group of individuals, even if they are not allocated power by political institutions, for example as specified in the constitution, may nonetheless possess political power. Namely, they can revolt, use arms, hire mercenaries, co-opt the military, or use economically costly but largely peaceful protests in order to impose their wishes on society. We refer to this type of political power as de facto political power, which itself has two sources. First, it depends on the ability of the group in question to solve its collective action problem, i.e., to ensure that people act together, even when any individual may have an incentive to free ride. For example, peasants in the Middle Ages, who were given no political power by the constitution, could sometimes solve the collective action problem and undertake a revolt against the authorities. Second, the de facto power of a group depends on its economic resources, which determine both their ability to use (or misuse) existing political institutions and also their option to hire and use force against different groups. Since we do not yet have a satisfactory theory of when groups are able to solve their collective action problems, our focus will be on the second source of de facto political power, hence: distribution of resourcest ⇒ de facto political powert . 6. This brings us to the evolution of one of the two main state variables in our framework, political institutions (the other state variable is the distribution of resources, including distribution of physical and human capital stocks, etc.). Political institutions and the distribution of resources are the state variables in this dynamic system because they typically change relatively slowly, and more importantly, they determine economic institutions and economic performance both directly and indirectly. Their direct effect is straightforward to understand. If political institutions place all political power in the hands of a single individual or a small group, economic institutions that provide protection of property rights and equal opportunity for the rest of the population are difficult to sustain. The indirect effect works through the channels discussed above: political institutions determine the distribution of de jure political power, which in turn affects the choice of economic institutions. This framework therefore introduces a natural concept of a hierarchy of institutions, with political institutions influencing equilibrium economic institutions, which then determine economic outcomes. Political institutions, though slow changing, are also endogenous. Societies transition from dictatorship to democracy, and change their constitutions to modify the constraints on power holders. Since, like economic institutions, political institutions are collective
392
D. Acemoglu et al.
choices, the distribution of political power in society is the key determinant of their evolution. This creates a tendency for persistence: political institutions allocate de jure political power, and those who hold political power influence the evolution of political institutions, and they will generally opt to maintain the political institutions that give them political power. However, de facto political power occasionally creates changes in political institutions. While these changes are sometimes discontinuous, for example when an imbalance of power leads to a revolution or the threat of revolution leads to major reforms in political institutions, often they simply influence the way existing political institutions function, for example, whether the rules laid down in a particular constitution are respected as in most functioning democracies, or ignored as in currentday Zimbabwe. Summarizing this discussion, we have: political powert ⇒ political institutionst+1 . Putting all these pieces together, a schematic (and simplistic) representation of our framework is as follows: economic de jure political performance ⇒ political t economic institutionst & powert ⇒ ⇒ institutionst & distribution of resourcest+1 de facto distribution political ⇒ political ⇒ of resourcest institutionst+1 powert The two state variables are political institutions and the distribution of resources, and the knowledge of these two variables at time t is sufficient to determine all the other variables in the system. While political institutions determine the distribution of de jure political power in society, the distribution of resources influences the distribution of de facto political power at time t. These two sources of political power, in turn, affect the choice of economic institutions and influence the future evolution of political institutions. Economic institutions determine economic outcomes, including the aggregate growth rate of the economy and the distribution of resources at time t + 1. Although economic institutions are the essential factor shaping economic outcomes, they are themselves endogenous and determined by political institutions and distribution of resources in society. There are two sources of persistence in the behavior of the system: first, political institutions are durable, and typically, a sufficiently large change in the distribution of political power is necessary to cause a change in political institutions, such as a transition from dictatorship to democracy. Second, when a particular group is rich relative to others, this will increase its de facto political power and enable it to push for economic and political institutions favorable to its interests. This will tend to reproduce the initial relative wealth disparity in the future. Despite these tendencies for persistence, the framework also emphasizes the potential for change. In particular, “shocks”, including changes in technologies and the international environment, that modify the balance of
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
393
(de facto) political power in society and can lead to major changes in political institutions and therefore in economic institutions and economic growth. A brief example might be useful to clarify these notions before commenting on some of the underlying assumptions and discussing comparative statics. Consider the development of property rights in Europe during the Middle Ages. There is no doubt that lack of property rights for landowners, merchants and proto-industrialists was detrimental to economic growth during this epoch. Since political institutions at the time placed political power in the hands of kings and various types of hereditary monarchies, such rights were largely decided by these monarchs. Unfortunately for economic growth, while monarchs had every incentive to protect their own property rights, they did not generally enforce the property rights of others. On the contrary, monarchs often used their powers to expropriate producers, impose arbitrary taxation, renege on their debts, and allocate the productive resources of society to their allies in return for economic benefits or political support. Consequently, economic institutions during the Middle Ages provided little incentive to invest in land, physical or human capital, or technology, and failed to foster economic growth. These economic institutions also ensured that the monarchs controlled a large fraction of the economic resources in society, solidifying their political power and ensuring the continuation of the political regime. The seventeenth century, however, witnessed major changes in the economic and political institutions that paved the way for the development of property rights and limits on monarchs’ power, especially in England after the Civil War of 1642 and the Glorious Revolution of 1688, and in the Netherlands after the Dutch Revolt against the Hapsburgs. How did these major institutional changes take place? In England, for example, until the sixteenth century the king also possessed a substantial amount of de facto political power, and leaving aside civil wars related to royal succession, no other social group could amass sufficient de facto political power to challenge the king. But changes in the English land market [Tawney (1941)] and the expansion of Atlantic trade in the sixteenth and seventeenth centuries [Acemoglu, Johnson and Robinson (2005)] gradually increased the economic fortunes, and consequently the de facto power of landowners and merchants. These groups were diverse, but contained important elements that perceived themselves as having interests in conflict with those of the king: while the English kings were interested in predating against society to increase their tax incomes, the gentry and merchants were interested in strengthening their property rights. By the seventeenth century, the growing prosperity of the merchants and the gentry, based both on internal and overseas, especially Atlantic, trade, enabled them to field military forces capable of defeating the king. This de facto power overcame the Stuart monarchs in the Civil War and Glorious Revolution, and led to a change in political institutions that stripped the king of much of his previous power over policy. These changes in the distribution of political power led to major changes in economic institutions, strengthening the property rights of both land and capital owners and spurred a process of financial and commercial expansion. The consequence was rapid economic growth, culminating in the Industrial Revolution, and a very different distribution of economic resources from that in the Middle Ages.
394
D. Acemoglu et al.
It is worth returning at this point to two critical assumptions in our framework. First, why do the groups with conflicting interests not agree on the set of economic institutions that maximize aggregate growth? So in the case of the conflict between the monarchy and the merchants, why does the monarchy not set up secure property rights to encourage economic growth and tax some of the benefits? Second, why do groups with political power want to change political institutions in their favor? For instance, in the context of the example above, why did the gentry and merchants use their de facto political power to change political institutions rather than simply implement the policies they wanted? The answers to both questions revolve around issues of commitment and go to the heart of our framework. The distribution of resources in society is an inherently conflictual, and therefore political, decision. As mentioned above, this leads to major commitment problems, since groups with political power cannot commit to not using their power to change the distribution of resources in their favor. For example, economic institutions that increased the security of property rights for land and capital owners during the Middle Ages would not have been credible as long as the monarch monopolized political power. He could promise to respect property rights, but then at some point, renege on his promise, as exemplified by the numerous financial defaults by medieval kings [e.g., Veitch (1986)]. Credible secure property rights necessitated a reduction in the political power of the monarch. Although these more secure property rights would foster economic growth, they were not appealing to the monarchs who would lose their rents from predation and expropriation as well as various other privileges associated with their monopoly of political power. This is why the institutional changes in England as a result of the Glorious Revolution were not simply conceded by the Stuart kings. James II had to be deposed for the changes to take place. The reason why political power is often used to change political institutions is related. In a dynamic world, individuals care not only about economic outcomes today but also in the future. In the example above, the gentry and merchants were interested in their profits and therefore in the security of their property rights, not only in the present but also in the future. Therefore, they would have liked to use their (de facto) political power to secure benefits in the future as well as the present. However, commitment to future allocations (or economic institutions) was not possible because decisions in the future would be decided by those who had political power in the future with little reference to past promises. If the gentry and merchants would have been sure to maintain their de facto political power, this would not have been a problem. However, de facto political power is often transient, for example because the collective action problems that are solved to amass this power are likely to resurface in the future, or other groups, especially those controlling de jure power, can become stronger in the future. Therefore, any change in policies and economic institutions that relies purely on de facto political power is likely to be reversed in the future. In addition, many revolutions are followed by conflict within the revolutionaries. Recognizing this, the English gentry and merchants strove not just to change economic institutions in their favor following their victories against the Stuart monarchy, but also to alter political institutions and the
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
395
future allocation of de jure power. Using political power to change political institutions then emerges as a useful strategy to make gains more durable. The framework that we propose, therefore, emphasizes the importance of political institutions, and changes in political institutions, as a way of manipulating future political power, and thus indirectly shaping future, as well as present, economic institutions and outcomes. This framework, though abstract and highly simple, enables us to provide some preliminary answers to our main question: why do some societies choose “good economic institutions”? At this point, we need to be more specific about what good economic institutions are. A danger we would like to avoid is that we define good economic institutions as those that generate economic growth, potentially leading to a tautology. This danger arises because a given set of economic institutions may be relatively good during some periods and bad during others. For example, a set of economic institutions that protects the property rights of a small elite might not be inimical to economic growth when all major investment opportunities are in the hands of this elite, but could be very harmful when investments and participation by other groups are important for economic growth [see Acemoglu (2003b)]. To avoid such a tautology and to simplify and focus the discussion, throughout we think of good economic institutions as those that provide security of property rights and relatively equal access to economic resources to a broad cross-section of society. Although this definition is far from requiring equality of opportunity in society, it implies that societies where only a very small fraction of the population have well-enforced property rights do not have good economic institutions. Consequently, as we will see in some of the historical cases discussed below, a given set of economic institutions may have very different implications for economic growth depending on the technological possibilities and opportunities. Given this definition of good economic institutions as providing secure property rights for a broad cross-section of society, our framework leads to a number of important comparative statics, and thus to an answer to our basic question. First, political institutions that place checks on those who hold political power, for example, by creating a balance of power in society, are useful for the emergence of good economic institutions. This result is intuitive; without checks on political power, power holders are more likely to opt for a set of economic institutions that are beneficial for themselves and detrimental for the rest of society, which will typically fail to protect property rights of a broad cross-section of people. Second, good economic institutions are more likely to arise when political power is in the hands of a relatively broad group with significant investment opportunities. The reason for this result is that, everything else equal, in this case power holders will themselves benefit from secure property rights.2 Third, good economic institutions are more likely to arise and persist when there are only limited rents that power holders can extract from the rest of society, since such rents would
2 The reason why we inserted the caveat of “a relatively broad group” is that when a small group with significant investment opportunities holds power, they may sometimes opt for an oligarchic system where their own property rights are protected, but those of others are not [see Acemoglu (2003b)].
396
D. Acemoglu et al.
encourage them to opt for a set of economic institutions that make the expropriation of others possible. These comparative statics therefore place political institutions at the center of the story, as emphasized by our term “hierarchy of institutions” above. Political institutions are essential both because they determine the constraints on the use of (de facto and de jure) political power and also which groups hold de jure political power in society. We will see below how these comparative statics help us understand institutional differences across countries and over time in a number of important historical examples. 1.3. Outline In the next section we discuss how economic institutions constitute the basis for a fundamental theory of growth, and we contrast this with other potential fundamental theories. In Section 3 we consider some empirical evidence that suggests a key role for economic institutions in determining long-run growth. We also emphasize some of the key problems involved in establishing a causal relationship between economic institutions and growth. We then show in Section 4 how the experience of European colonialism can be used as a ‘natural experiment’ which can address these problems. Having established the central causal role of economic institutions and their importance relative to other factors in cross-country differences in economic performance, the rest of the paper focuses on developing a theory of economic institutions. Section 5 discusses four types of explanation for why countries have different institutions, and argues that the most plausible is the social conflict view. According to this theory, bad institutions arise because the groups with political power benefit from bad institutions. The emphasis on social conflict arises naturally from our observation above that economic institutions influence the distribution of resources as well as efficiency. Different groups or individuals will therefore prefer different institutions and conflict will arise as each tries to get their own way. Section 6 delves deeper into questions of efficiency and asks why a political version of the Coase Theorem does not hold. We emphasize the idea that commitment problems are intrinsic to the exercise of political power. In Section 7 we argue that a series of historical examples of diverging economic institutions are best explained by the social conflict view. These examples illustrate how economic institutions are determined by the distribution of political power, and how this distribution is influenced by political institutions. Section 8 puts these ideas together to build our theory of institutions. In Section 9 we then consider two more extended examples of the theory in action, the rise of constitutional rule in early modern Europe, and the creation of mass democracy, particularly in Britain, in the nineteenth and twentieth centuries. Section 10 concludes with a discussion of where this research program can go next. 2. Fundamental causes of income differences We begin by taking a step back. The presumption in the introduction was that economic institutions matter, and should in fact be thought of as one of the key fundamental causes
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
397
of economic growth and cross-country differences in economic performance. How do we know this? 2.1. Three fundamental causes If standard economic models of factor accumulation and endogenous technical change only provide proximate explanations of comparative growth, what types of explanations would constitute fundamental ones? Though there is no conventional wisdom on this, we can distinguish three such theories: the first set of theories, our main focus in this chapter, emphasize the importance of economic institutions, which influence economic outcomes by shaping economic incentives; the second emphasize geography, and the third emphasize the importance of culture (a fourth possibility is that differences are due to “luck”, some societies were just lucky; however we do not believe that differences in luck by themselves constitute a sufficient fundamental causes of cross-country income differences). 2.1.1. Economic institutions At its core, the hypothesis that differences in economic institutions are the fundamental cause of different patterns of economic growth is based on the notion that it is the way that humans themselves decide to organize their societies that determines whether or not they prosper. Some ways of organizing societies encourage people to innovate, to take risks, to save for the future, to find better ways of doing things, to learn and educate themselves, solve problems of collective action and provide public goods. Others do not. The idea that the prosperity of a society depends on its economic institutions goes back at least to Adam Smith, for example in his discussions of mercantilism and the role of markets, and was prominent in the work of many nineteenth century scholars such as John Stuart Mill [see the discussion in Jones (1981)]: societies are economically successful when they have ‘good’ economic institutions and it is these institutions that are the cause of prosperity. We can think of these good economic institutions as consisting of an inter-related cluster of things. There must be enforcement of property rights for a broad cross-section of society so that all individuals have an incentive to invest, innovate and take part in economic activity. There must also be some degree of equality of opportunity in society, including such things as equality before the law, so that those with good investment opportunities can take advantage of them.3 One could think of other types of economic institutions and many explanations for growth and development have moved beyond models based on preferences, technology and factor endowments to focus on what might loosely be called institutions. One 3 In Acemoglu, Johnson and Robinson (2001), we coined the term institutions of private property for a cluster of good economic institutions, including the rule of law and the enforcement of property rights, and the term extractive institutions to designate institutions under which the rule of law and property rights are absent for large majorities of the population.
398
D. Acemoglu et al.
set of ideas, important for our work, has emphasized that conflict over resources and predation, as well as production, are fundamental forces in society. Scholars such as Skaperdas (1992), Grossman and Kim (1995, 1996), Hirshleifer (2001) and Dixit (2004) have examined how stable property rights can emerge in such circumstances. These scholars have studied almost institution free models and asked how the type of social order that underlies standard economic models might emerge endogenously. Closely related to this work is the research that shows how rent-seeking and redistributional conflict more generally has important implications for growth [e.g., Tornell and Velasco (1992), Murphy, Shleifer and Vishny (1991) Acemoglu (1995), Alesina and Perotti (1996), Benhabib and Rustichini (1996)]. Another literature, following in the footsteps of traditional accounts of economic growth by historians, following the lead of Adam Smith, has emphasized the perfection and spread of markets, clearly a key economic institution [Pirenne (1937), Hicks (1969)]. Problems of the imperfection or absence of markets can clearly have important ramifications for resource allocation, incentives and growth. A central role here has been played by capital markets. For example, Banerjee and Newman (1993) and Galor and Zeira (1993) propose canonical models of how imperfect financial markets can impede growth and development. Models of poverty traps in the tradition of Rosenstein-Rodan (1943), Murphy, Vishny and Shleifer (1989a, 1989b) and Acemoglu (1995, 1997), are based on the idea that market imperfections can lead to the existence of multiple Pareto-ranked equilibria. As a consequence a country can get stuck in a Pareto inferior equilibrium, associated with poverty, but getting out of such a trap necessitates coordinated activities that the market cannot deliver. Other mechanisms, such as increasing returns to scale, can lead to similar situations [e.g., Durlauf (1993), Krugman and Venables (1995), see Azariadis and Stachurski (2005), for other mechanisms and examples]. The implications of many other types of market imperfections have been considered, for example in the labor market [Aghion and Howitt (1994), Pissarides (2000)] and other scholars have examined the implications of industrial organization, market structure and the nature of competition [e.g., Acemoglu and Zilibotti (1997), Aghion et al. (2001), Aghion and Howitt (2005)]. The idea that market imperfections and economic institutions play a central role in development has also been important in the academic literature on development economics since its initiation. Both Adam Smith and Alfred Marshall argued that sharecropping was an inefficient way of organizing agriculture because it gave incorrect incentives to tenants. This argument has been formalized, and at the heart of a large literature on development are imperfections in tenancy, labor, land and credit markets [see Ray (1998), Bardhan and Udry (1999), Banerjee and Duflo (2005)]. Finally, the literature that one might broadly class as institutional has extensively discussed political economy models. Most influential is the early work of Perotti (1993), Saint-Paul and Verdier (1993), Alesina and Rodrik (1994) and Persson and Tabellini (1994) who developed dynamic models to examine the effect of redistributive taxation on growth. There are now many models where political mechanisms and outcomes can have important influences on the growth rate [see Ades and Verdier (1996), Krusell and
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
399
Ríos-Rull (1999), Bourguignon and Verdier (2000) and other contributions which we discuss in the body of the paper]. At some level then there is a bewildering array of ideas connecting institutions, both economic and political, to growth and development. In this chapter however, as will already be apparent, we do not attempt to survey all of these theories. Rather, we attempt to develop a perspective on this topic which revolves around what we see as the key issues. From the empirical side this entails really establishing the causal role of institutions in development. From the theoretical side this involves emphasizing the importance of understanding why institutions differ across countries. From the perspective of this chapter the main problem with most of the existing research is the lack of comparative statics and the absence of a truly comparative focus. For instance, in the model of Grossman and Kim (1995) stable property rights may emerge as an equilibrium, but whether they do so or not depends on parameters in the fighting technology which are hard to interpret in reality. Most models of imperfect markets and multiple equilibria fail to provide explanations either for why markets are incomplete or imperfect, or for how some societies manage to get into good equilibria while others do not. To the extent that imperfect market are grounded in imperfections in information or possibilities for opportunism, one would like to know how and why these vary across countries in ways which are consistent with the basic facts about relative economic outcomes. We believe that the structure of markets is endogenous, and partly determined by property rights. Once individuals have secure property rights and there is equality of opportunity, the incentives will exist to create and improve markets (even though achieving perfect markets would be typically impossible). Thus we expect differences in markets to be an outcome of differing systems of property rights and political institutions, not unalterable characteristics responsible for cross-country differences in economic performance. This motivates our focus on economic institutions related to the enforcement of the property rights of a broad cross-section of society. There are some genuinely comparative studies in the literature. For example, Banerjee and Newman (1993), Alesina and Rodrik (1994) and Persson and Tabellini (1994) all point to differences in wealth distribution as the key to success or failure. We will discuss other such theories, for example those connected to legal origins [e.g., La Porta et al. (1998)] later. Nevertheless, these studies are very different from the approach we propose in this chapter. 2.1.2. Geography While institutional theories emphasize the importance of man-made factors shaping incentives, an alternative is to focus on the role of “nature”, that is, on the physical and geographical environment. In the context of understanding cross-country differences in economic performance, this approach emphasizes differences in geography, climate and ecology that determine both the preferences and the opportunity set of individual economic agents in different societies. We refer to this broad approach as the “geography
400
D. Acemoglu et al.
hypothesis”. There are at least three main versions of the geography hypothesis, each emphasizing a different mechanism for how geography affects prosperity. First, climate may be an important determinant of work effort, incentives, or even productivity. This idea dates back at least to the famous French philosopher, Montesquieu (1748), who wrote in his classic book The Spirit of the Laws: “The heat of the climate can be so excessive that the body there will be absolutely without strength. So, prostration will pass even to the spirit; no curiosity, no noble enterprise, no generous sentiment; inclinations will all be passive there; laziness there will be happiness”, and “People are . . . more vigorous in cold climates. The inhabitants of warm countries are, like old men, timorous; the people in cold countries are, like young men, brave.” One of the founders of modern economics Marshall is another prominent figure who emphasized the importance of climate, arguing: “vigor depends partly on race qualities: but these, so far as they can be explained at all, seem to be chiefly due to climate” [Marshall (1890, p. 195)]. Second, geography may determine the technology available to a society, especially in agriculture. This view is developed by an early Nobel Prize winner in economics, Myrdal, who wrote “serious study of the problems of underdevelopment . . . should take into account the climate and its impacts on soil, vegetation, animals, humans and physical assets – in short, on living conditions in economic development” [Myrdal (1968, vol. 3, p. 2121)]. More recently, Diamond espouses this view, “. . . proximate factors behind Europe’s conquest of the Americas were the differences in all aspects of technology. These differences stemmed ultimately from Eurasia’s much longer history of densely populated . . . [societies dependent on food production]”, which was in turn determined by geographical differences between Europe and the Americas [Diamond (1997, p. 358)]. The economist Sachs has been a recent and forceful proponent of the importance of geography in agricultural productivity, stating that “By the start of the era of modern economic growth, if not much earlier, temperate-zone technologies were more productive than tropical-zone technologies . . .” [Sachs (2001, p. 2)]. The third variant of the geography hypothesis, especially popular over the past decade, links poverty in many areas of the world to their “disease burden”, emphasizing that: “The burden of infectious disease is similarly higher in the tropics than in the temperate zones” [Sachs (2000, p. 32)]. Bloom and Sachs (1998) claim that the prevalence of malaria, a disease which kills millions of children every year in sub-Saharan Africa, reduces the annual growth rate of sub-Saharan African economies by more than 1.3 percent a year (this is a large effect, implying that had malaria been eradicated in 1950, income per capita in sub-Saharan Africa would be double what it is today). 2.1.3. Culture The final fundamental explanation for economic growth emphasizes the idea that different societies (or perhaps different races or ethnic groups) have different cultures, because of different shared experiences or different religions. Culture is viewed as a
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
401
key determinant of the values, preferences and beliefs of individuals and societies and, the argument goes, these differences play a key role in shaping economic performance. At some level, culture can be thought to influence equilibrium outcomes for a given set of institutions. Possibly there are multiple equilibria connected with any set of institutions and differences in culture mean that different societies will coordinate on different equilibria. Alternatively, as argued by Greif (1994), different cultures generate different sets of beliefs about how people behave and this can alter the set of equilibria for a given specification of institutions (for example, some beliefs will allow punishment strategies to be used whereas others will not). The most famous link between culture and economic development is that proposed by Weber (1930) who argued that the origins of industrialization in western Europe could be traced to the Protestant reformation and particularly the rise of Calvinism. In his view, the set of beliefs about the world that was intrinsic to Protestantism were crucial to the development of capitalism. Protestantism emphasized the idea of predestination in the sense that some individuals were ‘chosen’ while others were not. “We know that a part of humanity is saved, the rest damned. To assume that human merit or guilt play a part in determining this destiny would be to think of God’s absolutely free decrees, which have been settled from eternity, as subject to change by human influence, an impossible contradiction” [Weber (1930, p. 60)]. But who had been chosen and who not? Calvin did not explain this. Weber (1930, p. 66) notes “Quite naturally this attitude was impossible for his followers . . . for the broad mass of ordinary men . . . So wherever the doctrine of predestination was held, the question could not be suppressed whether there was any infallible criteria by which membership of the electi could be known”. Practical solutions to this problem were quickly developed, “. . . in order to attain that self-confidence intense worldly activity is recommended as the most suitable means. It and it alone disperses religious doubts and gives the certainly of grace” [Weber (1930, pp. 66–67)]. Thus “however useless good works might be as a means of attaining salvation . . . nevertheless, they are indispensable as a sign of election. They are the technical means, not of purchasing salvation, but of getting rid of the fear of damnation” (p. 69). Though economic activity was encouraged, enjoying the fruits of such activity was not. “Waste of time is . . . the first and in principle the deadliest of sins. The span of human life is infinitely short and precious to make sure of one’s own election. Loss of time through sociability, idle talk, luxury, even more sleep than is necessary for health . . . is worthy of absolute moral condemnation . . . Unwillingness to work is symptomatic of the lack of grace” (pp. 104–105). Thus Protestantism led to a set of beliefs which emphasized hard work, thrift, saving, and where economic success was interpreted as consistent with (if not actually signaling) being chosen by God. Weber contrasted these characteristics of Protestantism with those of other religions, such as Catholicism, which he argued did not promote capitalism. For instance on his book on Indian religion he argued that the caste system blocked capitalist development [Weber (1958, p. 112)].
402
D. Acemoglu et al.
More recently, scholars, such as Landes (1998), have also argued that the origins of Western economic dominance are due to a particular set of beliefs about the world and how it could be transformed by human endeavor, which is again linked to religious differences. Although Barro and McCleary (2003) provide evidence of a positive correlation between the prevalence of religious beliefs, notably about hell and heaven, and economic growth, this evidence does not show a causal effect of religion on economic growth, since religious beliefs are endogenous both to economic outcomes and to other fundamental causes of income differences [points made by Tawney (1926), and Hill (1961b), in the context of Weber’s thesis]. Ideas about how culture may influence growth are not restricted to the role of religion. Within the literature trying to explain comparative development there have been arguments that there is something special about particular cultural endowments, usually linked to particular nation states. For instance, Latin America may be poor because of its Iberian heritage, while North America is prosperous because of its Anglo-Saxon heritage [Véliz (1994)]. In addition, a large literature in anthropology argues that societies may become ‘dysfunctional’ or ‘maladapted’ in the sense that they adopt a system of beliefs or ways or operating which do not promote the success or prosperity of the society [see Edgerton (1992), for a survey of this literature]. The most famous version of such an argument is due to Banfield (1958) who argued that the poverty of Southern Italy was due to the fact that people had adopted a culture of “amoral familiarism” where they only trusted individuals of their own families and refused to cooperate or trust anyone else. This argument was revived in the extensive empirical study of Putnam, Leonardi and Nanetti (1993) who characterized such societies as lacking “social capital”. Although Putnam and others, for example, Knack and Keefer (1997) and Durlauf and Fafchamps (2004), document positive correlations between measures of social capital and various economic outcomes, there is no evidence of a causal effect, since, as with religious beliefs discussed above, measures of social capital are potentially endogenous.
3. Institutions matter We now argue that there is convincing empirical support for the hypothesis that differences in economic institutions, rather than geography or culture, cause differences in incomes per-capita. Consider first Figure 1. This shows the cross-country bivariate relationship between the log of GDP percapita in 1995 and a broad measure of property rights, “protection against expropriation risk”, averaged over the period 1985 to 1995. The data on economic institutions come from Political Risk Services, a private company which assesses the risk that investments will be expropriated in different countries. These data, first used by Knack and Keefer (1995) and subsequently by Hall and Jones (1999) and Acemoglu, Johnson and Robinson (2001, 2002) are imperfect as a measure of economic institutions, but the findings are robust to using other available measures of economic institutions. The scatter plot
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
403
Figure 1. Average protection against risk of expropriation 1985–95 and log GDP per capita 1995.
shows that countries with more secure property rights, i.e., better economic institutions, have higher average incomes. It is tempting to interpret Figure 1 as depicting a causal relationship (i.e., as establishing that secure property rights cause prosperity). Nevertheless, there are well-known problems with making such an inference. First, there could be reverse causation – perhaps only countries that are sufficiently wealthy can afford to enforce property rights. More importantly, there might be a problem of omitted variable bias. It could be something else, e.g., geography, that explains both why countries are poor and why they have insecure property rights. Thus if omitted factors determine institutions and incomes, we would spuriously infer the existence of a causal relationship between economic institutions and incomes when in fact no such relationship exists. Trying to estimate the relationship between institutions and prosperity using Ordinary Least Squares, as was done by Knack and Keefer (1995) and Barro (1997) could therefore result in biased regression coefficients. To further illustrate these potential identification problems, suppose that climate, or geography more generally, matters for economic performance. In fact, a simple scatterplot shows a positive association between latitude (the absolute value of distance from the equator) and income per capita. Montesquieu, however, not only claimed that warm climate makes people lazy and thus unproductive, but also unfit to be governed by democracy. He argued that despotism would be the political system in warm climates. Therefore, a potential explanation for the patterns we see in Figure 1 is that there is an omitted factor, geography, which explains both economic institutions and economic performance. Ignoring this potential third factor would lead to mistaken conclusions.
404
D. Acemoglu et al.
Figure 2. Latitude and log GDP per capita 1995.
Even if Montesquieu’s story appears both unrealistic and condescending to our modern sensibilities, the general point should be taken seriously: the relationship shown in Figure 1, and for that matter that shown in Figure 2, is not causal. As we pointed out in the context of the effect of religion or social capital on economic performance, these types of scatterplots, correlations, or their multidimensional version in OLS regressions, cannot establish causality. What can we do? The solution to these problems of inference is familiar in microeconometrics: find a source of variation in economic institutions that should have no effect on economic outcomes, or depending on the context, look for a natural experiment. As an example, consider first one of the clearest natural experiments for institutions. 3.1. The Korean experiment Until the end of World War II, Korea was under Japanese occupation. Korean independence came shortly after the Japanese Emperor Hirohito announced the Japanese surrender on August 15, 1945. After this date, Soviet forces entered Manchuria and North Korea and took over the control of these provinces from the Japanese. The major fear of the United States during this time period was the takeover of the entire Korean peninsular either by the Soviet Union or by communist forces under the control of the former guerrilla fighter, Kim Il Sung. U.S. authorities therefore supported the influential nationalist leader Syngman Rhee, who was in favor of separation rather than a united communist Korea. Elections in the South were held in May 1948, amidst a widespread
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
405
boycott by Koreans opposed to separation. The newly elected representatives proceeded to draft a new constitution and established the Republic of Korea to the south of the 38th parallel. The North became the Democratic People’s Republic of Korea, under the control of Kim II Sung. These two independent countries organized themselves in very different ways and adopted completely different sets of institutions. The North followed the model of Soviet socialism and the Chinese Revolution in abolishing private property of land and capital. Economic decisions were not mediated by the market, but by the communist state. The South instead maintained a system of private property and the government, especially after the rise to power of Park Chung Hee in 1961, attempted to use markets and private incentives in order to develop the economy. Before this “natural experiment” in institutional change, North and South Korea shared the same history and cultural roots. In fact, Korea exhibited an unparalleled degree of ethnic, linguistic, cultural, geographic and economic homogeneity. There are few geographic distinctions between the North and South, and both share the same disease environment. For example, the CIA Factbook describes the climate of North Korea as “temperate with rainfall concentrated in summer” and that of South Korea as “temperate, with rainfall heavier in summer than winter”. In terms of terrain North Korea is characterized as consisting of “mostly hills and mountains separated by deep, narrow valleys; coastal plains wide in west, discontinuous in east”, while South Korea is “mostly hills and mountains; wide coastal plains in west and south”. In terms of natural resources North Korea is better endowed with significant reserves of coal, lead, tungsten, zinc, graphite, magnesite, iron ore, copper, gold, pyrites, salt, fluorspar, hydropower. South Korea’s natural resources are “coal, tungsten, graphite, molybdenum, lead, hydropower potential”. Both countries share the same geographic possibilities in terms of access to markets and the cost of transportation. Other man-made initial economic conditions were also similar, and if anything, advantaged the North. For example, there was significant industrialization during the colonial period with the expansion of both Japanese and indigenous firms. Yet this development was concentrated more in the North than the South. For instance, the large Japanese zaibatsu of Noguchi, which accounted for one third of Japanese investment in Korea, was centered in the North. It built large hydroelectric plants, including the Suiho dam on the Yalu river, second in the world only to the Boulder dam on the Colorado river. It also created Nippon Chisso, the second largest chemical complex in the world that was taken over by the North Korean state. Finally, in Ch’ongjin North Korea also had the largest port on the Sea of Japan. All in all, despite some potential advantages for the North,4 Maddison (2001) estimates that at the time of separation, North and South Korea had approximately the same income per capita. We can therefore think of the splitting on the Koreas 50 years ago as a natural experiment that we can use to identify the causal influence of a particular dimension of
4 Such initial differences were probably eradicated by the intensive bombing campaign that the United States unleashed in the early 1950’s on North Korea [see Cumings (2004, Chapter 1)].
406
D. Acemoglu et al.
Figure 3. GDP per capita in North and South Korea, 1950–98.
institutions on prosperity. Korea was split into two, with the two halves organized in radically different ways, and with geography, culture and many other potential determinants of economic prosperity held fixed. Thus any differences in economic performance can plausibly be attributed to differences in institutions. Consistent with the hypothesis that it is institutional differences that drive comparative development, since separation, the two Koreas have experienced dramatically diverging paths of economic development (Figure 3). By the late 1960’s South Korea was transformed into one of the Asian “miracle” economies, experiencing one of the most rapid surges of economic prosperity in history while North Korea stagnated. By 2000 the level of income in South Korea was $16,100 while in North Korea it was only $1,000. By 2000 the South had become a member of the Organization of Economic Cooperation and Development, the rich nations club, while the North had a level of per-capita income about the same as a typical sub-Saharan African country. There is only one plausible explanation for the radically different economic experiences on the two Koreas after 1950: their very different institutions led to divergent economic outcomes. In this context, it is noteworthy that the two Koreas not only shared the same geography, but also the same culture. It is possible that Kim Il Sung and Communist Party members in the North believed that communist policies would be better for the country and the economy in the late 1940s. However, by the 1980s it was clear that the communist economic policies in the North were not working. The continued efforts of the leadership to cling to these policies and to power can only be explained by those leaders wishing to look after their own interests at the expense of the population at large. Bad institutions are therefore
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
407
kept in place, clearly not for the benefit of society as a whole, but for the benefit of the ruling elite, and this is a pattern we encounter in most cases of institutional failure that we discuss in detail below. However convincing on its own terms, the evidence from this natural experiment is not sufficient for the purposes of establishing the importance of economic institutions as the primary factor shaping cross-country differences in economic prosperity. First, this is only one case, and in the better-controlled experiments in the natural sciences, a relatively large sample is essential. Second, here we have an example of an extreme case, the difference between a market-oriented economy and a communist one. Few social scientists today would deny that a lengthy period of totalitarian centrally planned rule has significant economic costs. And yet, many might argue that differences in economic institutions among capitalist economies or among democracies are not the major factor leading to differences in their economic trajectories. To establish the major role of economic institutions in the prosperity and poverty of nations we need to look at a larger scale “natural experiment” in institutional divergence. 3.2. The colonial experiment The colonization of much of the world by Europeans provides such a large scale natural experiment. Beginning in the early fifteenth century and massively intensifying after 1492, Europeans conquered many other nations. The colonization experience transformed the institutions in many diverse lands conquered or controlled by Europeans. Most importantly, Europeans imposed very different sets of institutions in different parts of their global empire, as exemplified most sharply by the contrast to the economic institutions in the northeast of America to those in the plantation societies of the Caribbean. As a result, while geography was held constant, Europeans initiated large changes in economic institutions, in the social organization of different societies. We will now show that this experience provides evidence which conclusively establishes the central role of economic institutions in development. Given the importance of this material and the details we need to provide, we discuss the colonial experience in the next section. 4. The Reversal of Fortune The impact of European colonialism on economic institutions is perhaps most dramatically conveyed by a single fact – historical evidence shows that there has been a remarkable Reversal of Fortune in economic prosperity within former European colonies. Societies like the Mughals in India, and the Aztecs and the Incas in the Americas were among the richest civilizations in 1500, yet the nation states that now coincide with the boundaries of these empires are among the poorer societies of today. In contrast, countries occupying the territories of the less-developed civilizations in North America, New Zealand and Australia are now much richer than those in the lands of the Mughals, Aztecs and Incas.
408
D. Acemoglu et al.
4.1. The reversal among the former colonies The Reversal of Fortune is not confined to such comparisons. Using reasonable proxies for prosperity before modern times, we can show that it is a much more systematic phenomenon. Our proxies for income per capita in pre-industrial societies are urbanization rates and population density. Only societies with a certain level of productivity in agriculture and a relatively developed system of transport and commerce can sustain large urban centers and a dense population. Figure 4 shows the relationship between income per capita and urbanization (fraction of the population living in urban centers with greater than 5000 inhabitants) today, and demonstrates that in the current era there is a significant relationship between urbanization and prosperity. Naturally, high rates of urbanization do not mean that the majority of the population lived in prosperity. In fact, before the twentieth century urban areas were centers of poverty and ill health. Nevertheless, urbanization is a good proxy for average income per capita in society, which closely corresponds to the measure we are using to look at prosperity. Figures 5 and 6 show the relationship between income per capita today and urbanization rates and (log) population density in 1500 for the sample of European colonies.5
Figure 4. Urbanization in 1995 and log GDP per capita in 1995.
5 The sample includes the countries colonized by the Europeans between the 15th and the 19th centuries as part of their overseas expansion after the discovery of the New World and the rounding of the Cape of Good Hope. It therefore excludes Ireland, parts of the Russian Empire and also the Middle East and countries briefly controlled by European powers as U.N. Mondays during the 20th century.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
409
Figure 5. Urbanization in 1500 and log GDP per capita in 1995, among former European colonies.
Figure 6. Log population density in 1500 and log GDP per capita in 1995, among former European colonies.
410
D. Acemoglu et al.
Figure 7. Urbanization in 1000 and 1500, among non-colonies.
We pick 1500 since it is before European colonization had an effect on any of these societies. A strong negative relationship, indicating a reversal in the rankings in terms of economic prosperity between 1500 and today, is clear in both figures. In fact, the figures show that in 1500 the temperate areas were generally less prosperous than the tropical areas, but this pattern too was reversed by the twentieth century. The urbanization data for these figures come from Bairoch (1988), Bairoch, Batou and Chèvre (1988), Chandler (1987), and Eggimann (1999). The data on population density are from McEvedy and Jones (1978). Details and further results are in Acemoglu, Johnson and Robinson (2002). There is something extraordinary about this reversal. For example, after the initial spread of agriculture there was remarkable persistence in urbanization and population density for all countries, including those which were to be subsequently colonized by Europeans. In Figures 7 and 8 we show the relationships for urbanization plotting separately the relationship between urbanization in 1000 and in 1500 for the samples of colonies and all other countries. Both figures show persistence, not reversal. Although Ancient Egypt, Athens, Rome, Carthage and other empires rose and fell, what these pictures show is that there was remarkable persistence in the prosperity of regions. Moreover, reversal was not the general pattern in the world after 1500. Figure 9 shows that within countries not colonized by Europeans in the early modern and modern period, there was no reversal between 1500 and 1995. There is therefore no reason to think that what is going on in Figures 5 and 6 is some sort of natural reversion to the mean.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
Figure 8. Urbanization in 1000 and 1500, among former European colonies.
Figure 9. Urbanization in 1500 and log GDP per capita in 1995, among non-colonies.
411
412
D. Acemoglu et al.
Figure 10. Evolution of urbanization among former European colonies.
4.2. Timing of the reversal When did the reversal occur? One possibility is that it arose shortly after the conquest of societies by Europeans but Figures 10 and 11 show that the previously-poor colonies surpassed the former highly-urbanized colonies starting in the late eighteenth and early nineteenth centuries, and this went hand in hand with industrialization. Figure 10 shows average urbanization in colonies with relatively low and high urbanization in 1500. The initially high-urbanization countries have higher levels of urbanization and prosperity until around 1800. At that time the initially low-urbanization countries start to grow much more rapidly and a prolonged period of divergence begins. Figure 11 shows industrial production per capita in a number of countries. Although not easy to see in the figure, there was more industry (per capita and total) in India in 1750 than in the United States. By 1860, the United States and British colonies with relatively good economic institutions, such as Australia and New Zealand, began to move ahead rapidly, and by 1953, a huge gap had opened up. 4.3. Interpreting the reversal Which of the three broad hypotheses about the sources of cross-country income differences are consistent with the reversal and its timing? These patterns are clearly
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
413
Figure 11. Evolution of industrial production per capita among former European colonies.
inconsistent with simple geography based views of relative prosperity. In 1500 it was the countries in the tropics which were relatively prosperous, in 2003 it is the reverse. This makes it implausible to base a theory of relative prosperity today, as Sachs (2000, 2001) does, on the intrinsic poverty of the tropics. This argument is inconsistent with the historical evidence. Nevertheless, following Diamond (1997), one could propose what Acemoglu, Johnson and Robinson (2002) call a “sophisticated geography hypothesis” which claims that geography matters but in a time varying way. For example, Europeans created “latitude specific” technology, such as heavy metal ploughs, that only worked in temperate latitudes and not with tropical soils. Thus when Europe conquered most of the world after 1492, they introduced specific technologies that functioned in some places (the United States, Argentina, Australia) but not others (Peru, Mexico, West Africa). However, the timing of the reversal, coming as it does in the nineteenth century, is inconsistent with the most natural types of sophisticated geography hypotheses. Europeans may have had latitude specific technologies, but the timing implies that these technologies must have been industrial, not agricultural, and it is difficult to see why industrial technologies do not function in the tropics (and in fact, they have functioned quite successfully in tropical Singapore and Hong Kong).6 6 A possible link is that proposed by Lewis (1978) who argued that tropical agriculture is less productive than temperate agriculture, and that an ‘agricultural revolution’ is a prerequisite to an industrial revolution because
414
D. Acemoglu et al.
Similar considerations weigh against the culture hypothesis. Although culture is slow-changing the colonial experiment was sufficiently radical to have caused major changes in the cultures of many countries that fell under European rule. In addition, the destruction of many indigenous populations and immigration from Europe are likely to have created new cultures or at least modified existing cultures in major ways [see Vargas Llosa (1989), for a fictionalized account of just such a cultural change]. Nevertheless, the culture hypothesis does not provide a natural explanation for the reversal, and has nothing to say on the timing of the reversal. Moreover, we discuss below how econometric models that control for the effect of institutions on income do not find any evidence of an effect of religion or culture on prosperity. The most natural explanation for the reversal comes from the institutions hypothesis, which we discuss next. 4.4. Economic institutions and the reversal Is the Reversal of Fortune consistent with a dominant role for economic institutions in comparative development? The answer is yes. In fact, once we recognize the variation in economic institutions created by colonization, we see that the Reversal of Fortune is exactly what the institutions hypothesis predicts. In Acemoglu, Johnson and Robinson (2002) we tested the connection between initial population density, urbanization, and the creation of good economic institutions. We showed that, others things equal, the higher the initial population density or the greater initial urbanization, the worse were subsequent institutions, including both institutions right after independence and today. Figures 12 and 13 show these relationships using the same measure of current economic institutions used in Figure 1, protection against expropriation risk today. They document that the relatively densely settled and highly urbanized colonies ended up with worse (or ‘extractive’) institutions, while sparselysettled and non-urbanized areas received an influx of European migrants and developed institutions protecting the property rights of a broad cross-section of society. European colonialism therefore led to an institutional reversal, in the sense that the previouslyricher and more-densely settled places ended up with worse institutions.7 To be fair, it is possible that the Europeans did not actively introduce institutions discouraging economic progress in many of these places, but inherited them from previous civilizations there. The structure of the Mughal, Aztec and Inca empires were already very hierarchical with power concentrated in the hands of narrowly based ruling elites and structured to extract resources from the majority for the benefit of a minority. Often
high agricultural productivity is needed to stimulate the demand for industrial goods. Though obviously such an explanation is not relevant for explaining industrialization in Singapore or Hong Kong, it may be relevant in other places. 7 The institutional reversal does not mean that institutions were necessarily better in the previously more densely-settled areas (see next paragraph). It only implies a tendency for the relatively poorer and less denselysettled areas to end up with better institutions than previously-rich and more densely-settled areas.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
415
Figure 12. Urbanization in 1500 and average protection against risk of expropriation 1985–95.
Figure 13. Log population density in 1500 and average protection against risk of expropriation 1985–95.
416
D. Acemoglu et al.
Europeans simply took over these existing institutions. Whether this is so is secondary for our focus, however. What matters is that in densely-settled and relatively-developed places it was in the interests of Europeans to have institutions facilitating the extraction of resources thus not respecting the property rights of the majority, while in the sparsely-settled areas it was in their interests to develop institutions protecting property rights. These incentives led to an institutional reversal. The institutional reversal, combined with the institutions hypothesis, predicts the Reversal of Fortune: relatively rich places got relatively worse economic institutions, and if these institutions are important, we should see them become relatively poor over time. This is exactly what we find with the Reversal of Fortune. Moreover, the institutions hypothesis is consistent with the timing of the reversal. Recall that the institutions hypothesis links incentives to invest in physical and human capital and in technology to economic institutions, and argues that economic prosperity results from these investments. Therefore, economic institutions should become more important when there are major new investment opportunities. The opportunity to industrialize was the major investment opportunity of the nineteenth century. Countries that are rich today, both among the former European colonies and other countries, are those that industrialized successfully during this critical period. 4.5. Understanding the colonial experience The explanation for the reversal that emerges from our discussion so far is one in which the economic institutions in various colonies were shaped by Europeans to benefit themselves. Moreover, because conditions and endowments differed between colonies, Europeans consciously created different economic institutions, which persisted and continue to shape economic performance. Why did Europeans introduce better institutions in previously-poor and unsettled areas than in previously-rich and densely-settled areas? The answer to this question relates to the comparative statics of our theoretical framework. Leaving a full discussion to later, we can note a couple of obvious ideas. Europeans were more likely to introduce or maintain economic institutions facilitating the extraction of resources in areas where they would benefit from the extraction of resources. This typically meant areas controlled by a small group of Europeans, and areas offering resources to be extracted. These resources included gold and silver, valuable agricultural commodities such as sugar, but most importantly people. In places with a large indigenous population, Europeans could exploit the population, be it in the form of taxes, tributes or employment as forced labor in mines or plantations. This type of colonization was incompatible with institutions providing economic or civil rights to the majority of the population. Consequently, a more developed civilization and a denser population structure made it more profitable for the Europeans to introduce worse economic institutions. In contrast, in places with little to extract, and in sparsely-settled places where the Europeans themselves became the majority of the population, it was in their interests to introduce economic institutions protecting their own property rights.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
417
4.6. Settlements, mortality and development The initial conditions we have emphasized so far refer to indigenous population density and urbanization. In addition, the disease environments differed markedly among the colonies, with obvious consequences on the attractiveness of European settlement. As we noted above, when Europeans settled, they established institutions that they themselves had to live under. Therefore, whether Europeans could settle or not had an exogenous effect on the subsequent path of institutional development. In other words, if the disease environment 200 or more years ago affects outcomes today only through its effect on institutions today, then we can use this historical disease environment as an exogenous source of variation in current institutions. From an econometric point of view we have a valid instrument which will enable us to pin down the casual effect of economic institutions on prosperity.8 We developed this argument in Acemoglu, Johnson and Robinson (2001) and investigated it empirically. We used initial conditions in the European colonies, particularly data from Curtin (1989, 1998) and Gutierrez (1986) on the mortality rates faced by Europeans (primarily soldiers, sailors, and bishops), as instruments for current economic institutions. The justification for this is that, outside of its effect on economic institutions during the colonial period, historical European mortality has no impact on current income levels. Figures 14 and 15 give scatter plots of this data against contemporaneous economic institutions and GDP per-capita. The sample is countries which were colonized by Europeans in the early modern and modern periods and thus excludes, among others, China, Japan, Korea, Thailand. Figure 14 shows the very strong relationship between the historical mortality risk faced by Europeans and the current extent to which property rights are enforced. A bivariate regression has an R 2 of 0.26. It also shows that there were very large differences in European mortality. Countries such as Australia, New Zealand and the United States were very healthy with life expectancy typically greater than in Britain. On the other hand mortality was extremely high in Africa, India and South-East Asia. Differential mortality was largely due to tropical diseases such as malaria and yellow fever and at the time it was not understood how these diseases arose nor how they could be prevented or cured. In Acemoglu, Johnson and Robinson (2001) we showed, using European mortality as an instrument for the current enforcement of property rights, that most of the gap between rich and poor countries today is due to differences in economic institutions. More precisely, we showed (p. 1387) that if one took two typical – in the sense that they both lie on the regression line – countries with high and low expropriation risk, like Nigeria and Chile, then almost the entire difference in incomes per-capita between 8 Although European mortality is potentially correlated with indigenous mortality, which may determine income today, in practice local populations have developed much greater immunity to malaria and yellow fever. Thus the historical experience of European mortality is a valid instrument for institutional development. See Acemoglu, Johnson and Robinson (2001).
418
D. Acemoglu et al.
Figure 14. Log mortality of potential European settlers and average protection against risk of expropriation 1985–95.
Figure 15. Log mortality of potential European settlers and log GDP per capita in 1995.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
419
them could be explained by the differences in the security of property rights. We also presented regression evidence that showed that once the effect of economic institutions on GDP per-capita was properly controlled for, geographical variables, such as latitude, whether or not a country is land-locked and the current disease environment, have no explanatory power for current prosperity. These ideas and results provide an interpretation of why there are strong correlations between geographical variables such as latitude and income per-capita. Basically this is because Europeans did not have immunity to tropical diseases during the colonial period and thus settler colonies tended, other things equal, to be created in temperate latitudes. Thus the historical creation of economic institutions was correlated with latitude. Without considering the role of economic institutions it is easy to find a spurious relationship between latitude and income per-capita. However, once economic institutions are properly controlled for, these relationships go away. There is no causal effect of geography on prosperity today, though geography may have been important historically in shaping economic institutions. What about the role of culture? On the face of it, the Reversal of Fortune is consistent with cultural explanations of comparative growth. The Europeans not only brought new institutions, they also brought their own cultures. There seem to be three main ways to test this idea. First, cultures may be systematically related to the national identity of the colonizing power. For example, the British may have implanted a ‘good’ AngloSaxon culture into colonies such as Australia and the United States, while the Spanish may have condemned Latin America by endowing it with a Hispanic or Iberian culture [the academic literature is full of ideas like this, for recent versions see Véliz (1994), North, Summerhill and Weingast (2000) and Wiarda (2001)]. Second, following Landes (1998), Europeans may have had a culture, for example a work ethic or set of beliefs, which was uniquely propitious to prosperity. Finally, following Weber (1930), Europeans also brought different religions with different implications for prosperity. Such a hypothesis could explain why Latin America is relatively poor since its citizens are primarily Roman Catholic, while North America is relatively rich because its citizens are mostly Protestant. However, the econometric evidence in Acemoglu, Johnson and Robinson (2001) is not consistent with any these views. Once we control properly for the effects of economic institutions, neither the identity of the colonial power, nor the contemporary fraction of Europeans in the population, nor the proportions of the populations of various religions, are significant determinants of income per capita. These econometric results are supported by historical examples. For instance, with respect to the identity of the colonizing power, in the 17th century the Dutch had perhaps the best domestic economic institutions in the world but the colonies they created in South-East Asia ended up with institutions designed for the extraction of resources, providing little economic or civil rights to the indigenous population. It is also be clear that the British in no way simply re-created British institutions in their colonies. For example, by 1619 the North American colony of Virginia had a representative assembly with universal male suffrage, something that did not arrive in Britain
420
D. Acemoglu et al.
Figure 16. Log population density in 1500 and log GDP per capita in 1995, among former British colonies.
itself until 1919. Another telling example is due to Newton (1914) and Kupperman (1993), who showed that the Puritan colony in Providence Island in the Caribbean quickly became just like any other Caribbean slave colony despite the Puritanical inheritance. Although no Spanish colony has been as successful economically as British colonies such as the United States, it is also important to note that Britain had many unsuccessful colonies (in terms of per capita income), such as in Africa, India and Bangladesh. To emphasize that the culture or the religion of the colonizer was not at the root of the divergent economic performances of the colonies, Figure 16 shows the reversal among the British colonies (with population density in 1500 on the horizontal axis). Just as in Figure 6, there is a strong negative relationship between population density in 1500 and income per capita today. With respect to the role of Europeans, Singapore and Hong Kong are now two of the richest countries in the world, despite having negligible numbers of Europeans. Moreover, Argentina and Uruguay have higher proportions of people of European descent than the United States and Canada, but are much less rich. To further document this, Figure 17 shows a similar reversal of fortune for countries where the fraction of those with European descent in 1975 is less than 5 percent of the population. Overall, the evidence is not consistent with a major role of geography, religion or culture transmitted by the identity of the colonizer or the presence of Europeans. Instead,
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
421
Figure 17. Log population density in 1500 and log GDP per capita in 1995, among former European colonies with current population less than 5% of European descent.
differences in economic institutions appear to be the robust causal factor underlying the differences in income per capita across countries. Institutions are therefore the fundamental cause of income differences and long-run growth.
5. Why do institutions differ? We saw that economic institutions matter, indeed are central in determining relative prosperity. In terms of the different fundamental theories that we discussed, there is overwhelming support for the emphasis of North and Thomas on institutions, as opposed to alternative candidate explanations which emphasize geography or culture. Yet, as we discussed in the introduction, finding that differences in economic institutions can account for the preponderance of differences in per-capita income between countries creates as many questions as it answers. For example, why do countries have different economic institutions? If poor countries are poor because they have bad economic institutions why do they not change them to better institutions? In short, to explain the evidence presented in the last two sections we need a theory of economic institutions. The theory will help to explain the equilibrium set of economic institutions in a particular country and the comparative statics of this theory will help to explain why economic institutions differ across countries.
422
D. Acemoglu et al.
In the Introduction (Section 1.2), we began to develop such a theory based on social conflict over economic institutions. We have now substantiated the first point we made there, that economic institutions determine prosperity. We must now move to substantiate our second point, that economic institutions must be treated as endogenous and what which economic institutions emerge depends on the distribution of political power in society. This is a key step towards our theory of economic institutions. In the process of substantiating this point however it is useful to step back and discuss other alternative approaches to developing a theory of economic institutions. Broadly speaking, there are four main approaches to the question of why institutions differ across countries, one of which coincides with the approach we are proposing, the social conflict view. We next discuss each of these separately and our assessment as to whether they provide a satisfactory framework for thinking about differences in economic institutions [see Acemoglu (2003a) and Robinson (1998), for related surveys of some of these approaches]. We shall conclude that the approach we sketched in Section 1.2 is by far the most promising one. 5.1. The efficient institutions view – the Political Coase Theorem According to this view, societies will choose the economic institutions that are socially efficient. How this surplus will be distributed among different groups or agents does not affect the choice of economic institutions. We stress here that the concept of efficiency is stronger than simply Pareto Optimality; it is associated with surplus, wealth or output maximization. The underlying reasoning of this view comes from the Coase Theorem. Coase (1960) argued that when different economic parties could negotiate costlessly, they will be able to bargain to internalize potential externalities. A farmer, who suffers from the pollution created by a nearby factory, can pay the factory owner to reduce pollution. Similarly, if the current economic institutions benefit a certain group while creating a disproportionate cost for another, these two groups can negotiate to change the institutions. By doing so they will increase the size of the total surplus that they can divide between themselves, and they can then bargain over the distribution of this additional surplus. Many different versions of the efficient economic institutions view have been proposed. Indeed, assuming that existing economic institutions are efficient is a standard methodological approach of economists, i.e., observing an institution, one tries to understand what are the circumstances that lead it to be efficient. For instance, Demsetz (1967) argued that private property emerged from common property when land became sufficiently scarce and valuable that it was efficient to privatize it. More recently, Williamson’s (1985) research, as well as Coase’s (1937) earlier work and the more formal analysis by Grossman and Hart (1986), argues that the governance of firms or markets is such as to guarantee efficiency (given the underlying informational and contractual constraints). Williamson argued that firms emerged as an efficient response to contractual problems that plague markets, particularly the fact that there may be ex-post opportunism when individuals make relationship specific investments. Another famous
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
423
application of the efficient institutions view is due to North and Thomas (1973) who argued that feudal economic institutions, such as serfdom, were an efficient contract between serfs and lords. The lords provided a public good, protection, in exchange for the labor of the serfs on their lands. In this view, without a modern fiscal system this was an efficient way to organize this exchange. [See Townsend (1993), for a recent version of the idea that other economic institutions of Medieval Europe, such as the open field system, were efficient.] Williamson and North and Thomas do not specify how different parties will reach agreement to achieve efficient economic institutions, and this may be problematical in the sense that many economic institutions relevant for development are collective choices not individual bargains. There may therefore be free riding problems inherent in the creation of efficient economic institutions. Nevertheless, the underlying idea, articulated by Becker (1958) and Wittman (1989), is that, at least in democracies, competition among pressure groups and political parties will lead to efficient policies and collective choices. In their view, an inefficient economic institution cannot be stable because a political entrepreneur has an incentive to propose a better economic institution and with the extra surplus generated will be able to make himself more attractive to voters. The efficient institutions view regards the structure of political institutions or power as irrelevant. This may matter for the distribution of total surplus, but it will not matter for efficiency itself. The ‘efficient’ set of political institutions is therefore indeterminate. The notion that a Coasian logic applies in political life as well as in economics is referred to by Acemoglu (2003a) as the Political Coase Theorem. Although the intuition that individuals and groups will strive towards efficient economic outcomes is appealing, there are both theoretical and empirical limits to the Political Coase Theorem. First, as argued by Acemoglu (2003a) and further discussed below, in politics there is an inherent commitment problem, often making the Political Coase Theorem inapplicable. Second, the Political Coase Theorem does not take us very far in understanding the effect of economic (or indeed political) institutions on economic outcomes – in this view, economic institutions are chosen efficiently, and all societies have the best possible economic institutions given their needs and underlying structures; hence, with the Political Coase Theorem, economic institutions cannot be the fundamental cause of income differences. However, the empirical results we discussed above suggest a major role for such institutional differences. The only way to understand these patterns is to think of economic institutions varying for reasons other than the underlying needs of societies. In fact, the instrumental variables and natural experiment strategies we exploited above make use precisely of a source of variation unrelated to the underlying needs of societies. For example, South and North Korea did not adopt very different economic systems because they had different needs, but because different systems were imposed on them for other exogenous reasons. In sum, we need a framework for understanding why certain societies consistently end up with economic institutions that are not, from a social point of view, in their best interests. We need a framework other than the Political Coase Theorem.
424
D. Acemoglu et al.
5.2. The ideology view A second view is that economic institutions differ across countries because of ideological differences – because of the similarity between this and the previous view, Acemoglu (2003a) calls this the Modified Political Coase Theorem. According to this view, societies may choose different economic institutions, with very different implications, because they – or their leaders – disagree about what would be good for the society. According to this approach, there is sufficient uncertainty about the right economic institutions that well-meaning political actors differ about what’s good for their own people. Societies where the leaders or the electorate turn out to be right ex post are those that prosper. The important point is that, just as with the efficient institutions view, there are strong forces preventing the implementation of policies that are known to be bad for the society at large. Several theoretical models have developed related ideas. For example, Piketty (1995) examined a model where different people have different beliefs about how much effort is rewarded in society. If effort is not rewarded then taxation generates few distortions and agents with such beliefs prefer a high tax rate. On the other hand if one believes that effort is rewarded then low taxes are preferable. Piketty showed that dispersion of beliefs could create dispersion of preferences over tax rates, even if all agents had the same objective. Moreover, incorrect beliefs could be self-fulfilling and persist over time because different beliefs tend to generate information consistent with those beliefs. Romer (2003) also presents a model where voters have different beliefs and showed that if mistakes are correlated, then society can choose a socially inefficient outcome. These models show that if different societies have different beliefs about what is socially efficient they can rationally choose different economic institutions. Belief differences clearly do play a role in shaping policies and institutions. Several interesting examples of this come from the early experience of independence in former British colonies. For example, it is difficult to explain Julius Nyerere’s policies in Tanzania without some reference to his and other leading politicians’ beliefs about the desirability of a socialist society. It also appears true that in India the Fabian socialist beliefs of Jawaharlal Nehru were important in governing the initial direction that Indian economic policies took. Nevertheless, the scope of a theory of institutional divergence and comparative development based on ideology seems highly limited. Can we interpret the differences in institutional development across the European colonies or the divergence in the economic institutions and policies between the North and South of Korea as resulting from differences in beliefs? For example, could it be the case that while Rhee, Park, and other South Korean leaders believed in the superiority of capitalist institutions and private property rights enforcement, Kim Il Sung and Communist Party members in the North believed that communist policies would be better for the country? In the case of South versus North Korea, this is certainly a possibility. However, even if differences in beliefs could explain the divergence in economic institutions in the immediate aftermath of separation, by the 1980s it was clear that the communist economic
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
425
policies in the North were not working. The continued effort of the leadership to cling to these policies, and to power, can only be explained by leaders looking after their own interests at the expense of the population at large. Most likely, North Korean leaders, the Communist Party, and bureaucratic elites are prolonging the current system, which gives them greater economic and political returns than the alternative, even though they fully understand the costs that the system imposes on the North Korean people. Differences in colonial policies are even harder to explain on the basis of differences in ideology. British colonists established different economic institutions in very different parts of the world: in the Caribbean they set up plantation societies based on slavery, supported by highly oppressive economic institutions. In contrast, the economic institutions that developed in areas where the British settled, and where there was no large population of indigenous people to be captured and put to work, and where slavery could not be profitably used, such as northeastern United States, Canada, Australia and New Zealand, were very different. Moreover, differences in the incentives of the colonists in various colonies are easy to understand: when they did not settle, they were choosing economic institutions simply to extract resources from the native population. When they settled in large numbers, economic institutions and policies emerged in order to protect them in the future and encourage investment and prosperity. These considerations make us tend towards a view which emphasizes the actions of key economic and political agents that are taken rationally and in recognition of their consequences, not simply differences in beliefs. We do not deny that belief differences and ideology often play important roles but we do not believe that a satisfactory theory of institutional differences can be founded on differences in ideology. 5.3. The incidental institutions view The efficient institutions view is explicitly based on economic reasoning: the social costs and benefits of different economic institutions are weighed against each other to determine which economic institutions should prevail. Efficiency arises because individuals ultimately calculate according to social costs and benefits. Institutions are therefore choices. A different approach, popular among many political scientists and sociologists, but also some economists, is to downplay choices and to think of institutions, both economic and political, as the by-product or unintended consequence of other social interactions or historical accidents. In other words, historical accidents at critical junctures determine institutions, and these institutions persist for a long time, with significant consequences. Here, we discuss two such theories. The first is the theory of political institutions developed by Moore (1966) in his Social Origins of Dictatorship and Democracy, the second is the recent emphasis in the economics literature on legal origins, for example as in the work of Shleifer and his co-authors [La Porta et al. (1998, 1999), Djankov et al. (2002, 2003), Glaeser and Shleifer (2002)]. Moore attempted to explain the different paths of political development in Britain, Germany and Russia. In particular, he investigated why Britain evolved into a democ-
426
D. Acemoglu et al.
racy, while Germany succumbed to fascism and Russia had a communist revolution. Moore stressed the extent of commercialization of agriculture and resulting labor relations in the countryside, the strength of the ‘bourgeoisie’, and the nature of class coalitions. In his theory, democracy emerged when there was a strong, politically assertive, commercial middle class, and when agriculture had commercialized so that there were no feudal labor relations in the countryside. Fascism arose when the middle classes were weak and entered into a political coalition with landowners. Finally, a communist revolution resulted when the middle classes were non-existent, agriculture was not commercialized, and rural labor was repressed through feudal regulations. In Moore’s theory, therefore, class coalitions and the way agriculture is organized determine which political institutions will emerge. However, the organization of agriculture is not chosen with an eye to its effects on political institutions, so these institutions are an unintended consequence. Although Moore was not explicitly concerned with economic development, it is a direct implication of his analysis that societies may end up with institutions that do not maximize income or growth, for example, when they take the path to communist revolution. Beginning with the work on shareholder rights [La Porta et al. (1998)], continuing to the efficiency of government [La Porta et al. (1999)] and more recently the efficiency of the legal system [Djankov et al. (2003)], Shleifer and his co-authors have argued that a central source of variation in many critical economic institutions is the origin of the legal system. For example, “Civil laws give investors weaker legal rights than common laws do, independent of the level of per-capita income. Common-law countries give both shareholders and creditors – relatively speaking – the strongest, and French-civil-law countries the weakest, protection” [La Porta et al. (1998, p. 1116)]. These differences have important implications for resource allocation. For example, when shareholders have poor protection of their rights, ownership of shares tends to be more highly concentrated. Djankov et al. (2003) collected a cross-national dataset on how different countries legal systems dealt with the issue of evicting a tenant for nonpayment of rent and collecting on a bounced check. They used these data to construct an index of procedural formalism of dispute resolution for each country and showed that such formalism was systematically greater in civil than in common law countries, and is associated with higher expected duration of judicial proceedings, less consistency, less honesty, less fairness in judicial decisions, and more corruption. Legal origins therefore seems to matter for important institutional outcomes. Where do legal origins come from? The main argument is that they are historical accidents, mostly related to the incidence of European colonialism. For example, Latin American countries adopted the Napoleonic codes in the nineteenth century because these were more compatible with their Spanish legal heritage. Importantly, the fact that Latin American countries therefore have ‘French legal origin’ is due to a historical accident and can be treated as exogenous with respect to current institutional outcomes. What about the difference between common law and civil law? Glaeser and Shleifer (2002) argue that the divergence between these systems stems from the medieval period
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
427
and reflects the balance of power between the lords and the king in England and France. Once these systems established, they persisted long after the initial rationale vanished. Although we believe that historical accidents and persistence are important, in reality the aspect of choice over institutions seems too important to be denied. Even if institutions have a tendency to persist, their persistence is still a choice, in the sense that if the agents decided to change institutions, change would be possible. There are important examples from history of countries radically changing their legal systems such as in Japan after the Meiji restoration, Russia after the Crimean War, and Turkey under Mustafa Kemal in the 1920’s. Another example might be central planning of the economy. Though many countries adopted this way of organizing the economy some abandoned it while others, such as North Korea and Cuba, still maintain it. The point here is that though institutions may in some circumstances be the incidental outcome of history, at some point people will start to ask why society has the institutions that it does and to consider other alternatives. At this point we are back in the realm of choice. 5.4. The social conflict view According to this view, economic (and political) institutions are not always chosen by the whole society (and not for the benefit of the whole society), but by the groups that control political power at the time (perhaps as a result of conflict with other groups). These groups will choose the economic institutions that maximize their own rents, and the economic institutions that result may not coincide with those that maximize total surplus, wealth or income. For example, economic institutions that enforce property rights by restricting state predation may not be in the interest of a ruler who wants to appropriate assets in the future. By establishing property rights, this ruler would be reducing his own future rents, so may well prefer economic institutions other than enforced private property. Therefore, equilibrium economic institutions will not be those that maximize the size of the overall pie, but the slice of the pie taken by the powerful groups. The first systematic development of this point of view in the economics literature is North (1981), who argued in the chapter on “A Neoclassical Theory of the State” that agents who controlled the state should be modeled as self-interested. He then argued that the set of property rights that they would choose for society would be those that maximized their payoff and because of ‘transactions costs’, these would not necessarily be the set that maximized social welfare. One problem with North’s analysis is that he does not clarify what the transactions costs creating a divergence between the interests of the state and the citizens are. Here, we will argue that commitment problems are at the root of this divergence. The notion that elites, i.e., the politically powerful, may opt for economic institutions which increase their incomes, often at the expense of society, is of course also present in much of the Marxist and dependency theory literature. For example, Dobb (1948), Brenner (1976, 1982) and Hilton (1981) saw feudalism, contrary to North and Thomas’s (1973) model, as a set of institutions designed to extract rents from the peasants at the
428
D. Acemoglu et al.
expense of social welfare.9 Dependency theorists such as Williams (1944), Wallerstein (1974–1980), Rodney (1972), Frank (1978) and Cardosso and Faletto (1979) argued that the international trading system was designed to extract rents from developing countries to the benefit of developed countries. The social conflict view includes situations where economic institutions may initially be efficient for a set of circumstances but are no longer efficient once the environment changes. For example, Acemoglu, Aghion and Zilibotti (2002) show that though certain sorts of organizations may be useful for countries a long way from the technological frontier, it may be socially efficient to change them subsequently. This may not happen however because it is not privately rational. An interesting example may be the large business enterprises (the chaebol) of South Korea. In the context of political institutions, one might then develop a similar thesis. Certain sets of institutions are efficient for very poor countries but they continue to exist even after they cease to be the efficient institutional arrangement. In stark contrast to the efficient institutions view, political institutions play a crucial role in the social conflict view. Which economic institutions arise depends on who has political power to create or block different economic institutions. Since political institutions play a central role in the allocation of such power they will be an intimate part of a social conflict theory of economic institutions. What distinguishes the social conflict view from the ideological view is that social conflict can lead to choices of economic institutions which cause underdevelopment even when all agents have common knowledge that this is so. What distinguishes it from the incidental view is that it emphasizes that institutional choices which cause underdevelopment are conscious choices, rather than the result of some historical accident. The aspect that distinguishes the social conflict view from the efficient institutions view is that it does not assume that institutions are always efficient. This is one possible outcome but it is not the only one or indeed the most likely. Why is this? Why cannot efficiency be separated from distribution? We discuss this issue in the next section.
6. Sources of inefficiencies Having motivated our first two assertions in Section 1.2, we are now in a position to discuss the third, related to the importance of commitment problems. The inability to commit to how political power will be used in the future means that the impact of economic institutions on efficiency cannot be separated from their effects on distribution.10 In any market situation where economic exchange takes place, and the quid is separated from the pro quo, issues of commitment will arise. That these issues are of crucial
9 Postan (1966, pp. 603–604) famously estimated that lords extracted about 50% of the entire production of peasants. 10 An alternative approach would be to stress informational asymmetries [Farrell (1987)].
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
429
importance has been recognized in the literatures on incomplete contracts and renegotiation [e.g., Hart (1995)]. Nevertheless, if the legal system functions properly, there is an array of enforceable contracts that owners can sign with managers, workers with employers, borrowers with lenders, etc. These contracts can be enforced because there is an authority, a third party, with the power to enforce contracts. Although the authority that is delegated to enforce contracts and to resolve disputes varies depending on the exact situation, all such power ultimately emanates from the state, which, in modern society, has a near-monopoly on the use of legitimate coercion. An owner and manager can write a contract because they believe that the state, and its agents the courts, would be impartial enforcers of the contract. In contrast, if, for example, a manager believed that the state would be aligned with the interests of the owner and refuse to punish the owner if and when he failed to make a payment stipulated by the contract, then the contract would have little value. Therefore, the presence of an impartial enforcer is important for contracting. The problem when it comes to institutional choices is that there is no such impartial third party that can be trusted to enforce contracts. This is the origin of the commitment problem in politics.11 To elaborate on this point, let us consider a situation where society can be governed as a dictatorship or as a democracy. Imagine that the dictator does not relinquish his power, but instead he promises that he will obey the rules of democracy, so that individuals can undertake the same investments as they would in democracy. This promise would not necessarily be credible. As long as the political system remains a dictatorship, there is no higher authority to make the dictator stick to his promise. There is no equivalent of a contract that can be enforced by an impartial third-party. After all, the dictator has the monopoly of military and political power, so he is the final arbiter of conflicting interests. There is no other authority to force the dictator to abide by his promises. A similar problem plagues the reverse solution, whereby the dictator agrees to a voluntary transition to democracy in return for some transfers in the future to compensate him for the lost income and privileges. Those who will benefit from a transition to democracy would be willing to make such promises, but once the dictator relinquishes his political power, there is no guarantee that citizens would agree to tax themselves in order to make payments to this former dictator. Promises of compensation to a former dictator are typically not credible. The essence of the problem is commitment. Neither party can commit to compensate the other nor can they commit to take actions that would not be in their interests ex post. The reason why commitment problems are severe in these examples is because we are
11 Many scholars have emphasized the fact that a key feature of political economy is that there is no third party
which can enforce the promises made by the state and that this leads to problems of commitment and endemic inefficiencies. This idea is discussed by North (1990) and Olson (1993), is central to the work of Weingast (1997, 1998) and is implicit in many other studies. See also Grossman and Noh (1994), Dixit (1996), Dixit and Londregan (1995), Besley and Coate (1998) and Powell (2004) for discussions of how inability to commit generates inefficiencies in political outcomes.
430
D. Acemoglu et al.
dealing with political power. Different institutions are associated with different distributions of political power, and there is no outside impartial party with the will and the power to enforce agreements. In some cases, there may be self-enforcing promises that maintain an agreement. Acemoglu (2003a) discusses such possibilities, but in general, there are limits to such self-enforcing agreements, because they require the participants to be sufficiently patient, and when it comes to matters of political power, the future is uncertain enough that no party would behave in a highly patient manner. Based on this reasoning, we can now discuss three different channels via which the presence of commitment problems will lead to the choice and persistence of inefficient institutions. 6.1. Hold-up Imagine a situation in which an individual or a group holds unconstrained political power. Also suppose that productive investments can be undertaken by a group of citizens or producers that are distinct from the “political elites”, i.e., the current power holders. The producers will only undertake the productive investments if they expect to receive the benefits from their investments. Therefore, a set of economic institutions protecting their property rights are necessary for investment. Can the society opt for a set of economic institutions ensuring such secure property rights? The answer is often no (even assuming that “society” wants to do so). The problem is that the political elites – those in control of political power – cannot commit to respect the property rights of the producers once the investment are undertaken. Naturally, ex ante, before investments are undertaken, they would like to promise secure property rights. But the fact that the monopoly of political power in their hands implies that they cannot commit to not hold-up producers once the investments are sunk. This is an obvious parallel to the hold-up problem in the theory of the firm, where once one of the parties in a relationship has undertaken investments specific to the relationship, other parties can hold her up, and capture some of the returns from her investments. As in the theory of the firm, the prospect of hold-up discourages investment. But now the problem is much more severe, since it is not only investments that are specific to a relationship that are subject to hold-up, but all investments. This is therefore an example of how inefficient economic institutions arise because of a monopoly of political power. Those with political power cannot commit not to use their political power ex post, and this translates directly into a set of economic institutions that do not provide secure property rights to groups without political power. The consequence is clear: without such protection, productive investments are not undertaken, and opportunities for economic growth go unexploited. The reason why these inefficient economic institutions persist (or may be the equilibrium institutions of the society) is related to commitment problems. Parallel to our above example of inducing the dictator to relinquish power, there are two ways to introduce secure property rights. First, in principle, political elites could promise to respect property rights. However, mere promises would not be credible, unless backed up by
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
431
the political elites relinquishing power, and this would mean relinquishing their rents and privileges. Second, political elites can be bought off by the beneficiaries of a system of more secure property rights. This would typically be achieved by a promise of future payments. For example, after investments are undertaken and output is produced, a share can be given to the political elites. But, as pointed out above, there is another, reverse commitment problem here; the beneficiaries of the new regime cannot commit to make the promised payments to the previous political elites. Many real world examples illustrate the commitment problems involved in limiting the use of political power. In practice, although buying off dictators and persuading them to leave power is difficult, there have been many attempts to do so, usually by trying to guarantee that they will not be persecuted subsequently. One way of doing this is to give them asylum in another country. Nevertheless, such attempts rarely succeed, most likely again because of commitment problems (the new regime cannot commit to abide by its promises). An illustrative example of this is the attempts by the Reagan administration to persuade Jean-Claude (‘Baby Doc’) Duvalier to relinquish power in Haiti in 1986. In the face of a popular uprising and rising social and economic chaos, the Reagan administration, via the intermediation of the Jamaican Prime Minster Edward Seaga, tried to persuade Duvalier to go into exile. He at first agreed and the White House announced his departure on January 30th, but the next day he changed his mind, unsure that he would really be protected, and stayed in Haiti. One month later he was forced into exile in France by the military. A more common, and in many ways more interesting strategy to induce dictators to relinquish power is to try to structure political institutions so as to guarantee that they will not be punished. Such institutional changes are sometimes important in transitions to democracy. For example, President Pinochet was willing to abide by the results of the 1989 plebiscite he lost in Chile because as a senator the Constitution protected him from prosecution. It was only when he left the country that he was vulnerable. Although Pinochet’s experience illustrates an example of structuring political institutions to achieve commitment, to create durable institutions constraining future use of political power is difficult in practice. These difficulties are well illustrated by the transition from white rule in Rhodesia to majority rule in Zimbabwe. Facing an unwinable guerrilla war, the white elite in Rhodesia sought to negotiate a transition of majority rule, but with enough institutional safeguards that their rents would be protected. These safeguards included the electoral system they wanted, which was used for the first post-independence elections, and massive over-representation in parliament [Reynolds (1999, p. 163)]. Whites were guaranteed 20% of the seats in the legislature for seven years despite making up only 2–3% of the population and were guaranteed 10 seats of the 40 seat senate. Clauses of the 1980 Constitution were also aimed at directly guaranteeing the property rights of the whites. In particular land reform was outlawed for 10 years after which it could only take place if compensated. The white negotiators at the Lancaster House talks in 1979 that produced these agreements understood that any promises made by the black majority negotiators about what would happen after independence could not be believed. They sought therefore to find
432
D. Acemoglu et al.
a set of rules that would get around this problem [Herbst (1990, pp. 13–36)]. Nevertheless, these guarantees were not enough to protect the property rights (and rents) of the whites in anything other than the short run. The Mugabe regime quickly absorbed the other factions from among the African guerrilla opposition, and more moderate relatively pro-white groups, such as Abel Muzorewa’s United African National Council, crumbled. In 1985 the Mugabe regime switched back to the electoral system it preferred [Reynolds (1999, p. 164)] and in 1987, at the first possible opportunity, it removed the guaranteed representation for whites. Though in 1987 Mugabe nominated white candidates for these seats [Horowitz (1991, pp. 135–136)], this did not last for long. In 1990 the senate was abolished. Finally, in 1990 the Constitution was amended to allow for the redistribution of land. Since this time the Mugabe government has begun a sustained policy of land redistribution away from whites through legal and extra-legal means. 6.2. Political losers Another related source of inefficient economic institutions arises from the desire of political elites to protect their political power. Political power is the source of the incomes, rents, and privileges of the elite. If their political power were eroded, their rents would decline. Consequently, the political elite should evaluate every potential economic change not only according to its economic consequences, such as its effects on economic growth and income distribution, but also according to its political consequences. Any economic change that will erode the elite’s political power is likely to reduce their economic rents in the long run. As an example, imagine a change in economic institutions that will increase economic growth, but in doing so, will also enrich groups that could potentially contest political power in the future. Everything else equal, greater economic growth is good for those holding political power. It will create greater returns on the assets that they possess, and also greater incomes that they can tax or expropriate. However, if their potential enemies are enriched, this also means greater threats against their power in the future. Fearing these potential threats to their political power, the elites may oppose changes in economic institutions that would stimulate economic growth. That the threat of becoming a political loser impedes the adoption of better institutions is again due to a commitment problem. If those who gained political power from institutional change could promise to compensate those who lost power then there would be no incentive to block better institutions. There are many historical examples illustrating how the fear of losing political power has led various groups of political and economic elites to oppose institutional change and also the introduction of new technologies. Perhaps the best documented examples come from the attitude of the elites to industrialization during the nineteenth century [see Acemoglu and Robinson (2000b, 2002)]. There were large differences between the rates at which countries caught up with British industrialization with many countries completely failing to take advantage of the new technologies and opportunities. In most of these cases, the attitudes of political elites towards industrialization, new technology
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
433
and institutional change appear to have been the decisive factor, and these attitudes were driven by their fears of becoming political losers. These issues are best illustrated by the experiences of Russia and Austria-Hungary. In both Russia and Austria-Hungary, absolutist monarchies feared that promoting industrialization would undermine their political power. In Russia, during the reign of Nikolai I between 1825 and 1855 only one railway line was built in Russia, and this was simply to allow the court to travel between Moscow and St. Petersburg. Economic growth and the set of institutions that would have facilitated it were opposed since, as Mosse (1992, p. 19) puts it “it was understood that industrial development might lead to social and political change”. In a similar vein, Gregory (1991, p. 74) argues: “Prior to the about face in the 1850s, the Russian state feared that industrialization and modernization would concentrate revolution minded workers in cities, railways would give them mobility, and education would create opposition to the monarchy”. It was only after the defeat in the Crimean War that Nikolai’s successor, Alexandr II, initiated a large scale project of railway building and an attempt to modernize the economy by introducing a western legal system, decentralizing government, and ending feudalism by freeing the serfs. This period of industrialization witnessed heightened political tensions, consistent with the fears of the elites that times of rapid change would destabilize the political status quo and strengthen their opposition [McDaniel (1991) gives a detailed account of these events, see also Mosse (1958)]. The consensus view amongst historians also appears to be that the main explanation for the slow growth of Austria-Hungary in the nineteenth century was lack of technology adoption and institutional change, again driven by the opposition of the state to economic change. This view was proposed by Gerschenkron who argued that the state not only failed to promote industrialization, but rather, “economic progress began to be viewed with great suspicion and the railroads came to be regarded, not as welcome carriers of goods and persons, but as carriers of the dreaded revolution. Then the state clearly became an obstacle to the economic development of the country” (1970, p. 89). See also Gross (1973). The analysis of Freudenberger (1967, pp. 498–499) is similar. As with the Tsar, the Hapsburg emperors opposed the building of railways and infrastructure and there was no attempt to develop an effective educational system. Blum (1943) pointed to the pre-modern institutional inheritance as the major blockage to industrialization arguing (p. 26) that “these living forces of the traditional economic system were the greatest barrier to development. Their chief supporter was . . . Emperor Francis. He knew that the advances in the techniques of production threatened the life of the old order of which he was so determined a protector. Because of his unique position as final arbiter of all proposals for change he could stem the flood for a time. Thus when plans for the construction of a steam railroad were put before him, he refused to give consent to their execution ‘lest revolution might come into the country’.”
434
D. Acemoglu et al.
6.3. Economic losers A distinct but related source of inefficiency stems from the basic supposition of the social conflict view that different economic institutions imply different distributions of incomes. This implies that a move from a bad to a better set of economic institutions will make some people or groups worse off (and will not be Pareto improving). This in turn implies that such groups will have an incentive to block or impede such institutional changes even if they benefit the whole of society in some aggregate sense. The idea that economic losers impede the choice of efficient economic institutions and economic policies is widespread in economics and was seen earliest in the literature on international trade. Even though free trade may be socially desirable, individuals invested in sectors in which an economy does not enjoy comparative advantage will lose economically from free trade. Since at least the work of Schattschneider (1935) the role of economic losers has been central in understanding why free trade is not adopted. In the context of development economics, this idea was first discussed by Kuznets (1968), developed at length by Olson (1982, 2000) and Mokyr (1990), and formalized by Krusell and Ríos-Rull (1996) and Parente and Prescott (1999, 2005). Most of the examples discussed in the development literature on economic losers are about technological change – people with specific investments in obsolete technology try to block the introduction of better technology. The most celebrated example is the case of the Luddites, skilled weavers in early nineteenth century England who smashed new mechanized looms which threatened to lead to massive cuts in their wages [see Thomis (1970), Randall (1991)]. Scott (2000, p. 200) relates a similar example from modern Malaysia, “When, in 1976, combine harvesters began to make serious inroads into the wages of poor villagers, the entire region experienced a rash of machine-breaking and sabotage reminiscent of the 1830s in England”. That better economic institutions are blocked by individuals whose incomes are threatened by such change is again due to a problem of commitment. If those whose incomes rose when economic institutions changed could promise to compensate those whose incomes fell then there would be no incentive to block better economic institutions. Nevertheless, it is difficult to commit to such transfers. To consider again the example of the Luddites, the factory owners could have promised to pay the weavers high wages in the future even though their skills were redundant. Once the new technology was in place however, owners would have a clear incentive to fire the weavers and hire much cheaper unskilled workers.12 Although the problem of economic losers is appealing at first sight, has received some attention in the economics literature, and fits into our framework by emphasizing the importance of commitment problems, we view it both theoretically and empirically 12 One possible way round this problem would be for the owners, if they could afford it, to compensate the
weavers in advance for their lower future wages. But this would raise the reverse commitment problem: the weavers would have an incentive to take the money and still break the machines – i.e., they could not commit to not blocking the innovations that would reduce their wages even after they had taken the money.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
435
less important than the holdup and the political loser problems. First, as pointed out in Acemoglu and Robinson (2000b), in theories emphasizing issues of economic losers, there are implicit assumptions about politics, which, when spelled out, imply that political concerns must be important whenever issues of economic losers are present. The idea of economic losers is that certain groups, fearing that they will lose their economic rents, prevent adoption of beneficial economic institutions or technologies. The assumption in this scenario is that these groups have the political power to block socially beneficial changes. But then, if they have the political power to block change, why would not they allow the change to take place and then use their political power to redistribute some of the gains to themselves? The implicit assumption must therefore be that groups losing economically also experience a reduction in their political power, making it impossible for them to redistribute the gains to themselves after to change takes place. This reasoning therefore suggests that whether certain groups will lose economically or not is not as essential to their attitudes towards change as whether their political power will be eroded. Problems of political losers therefore seem much more important than problems of economic losers. Possibly for this reason, advocates of the economic losers view have been unable to come up with any well documented examples where the economic losers hypothesis can actually explain first-order patterns of development. For instance, while it is true that the Luddites tried to break machines, they singularly failed to halt the progress of agricultural technology in nineteenth century Britain. The same is true for Malaysia in the 1970s, one of the fastest growing economies in the world at that time. Neither set of workers had sufficient political power to stop change. Indeed, when political powerful groups became economic losers, such as landowners in nineteenth century England who saw land prices and agricultural rents fall rapidly after 1870, they did nothing to block change because their political power allowed them to benefit from efficient economic institutions [Acemoglu and Robinson (2002)]. Perhaps the most interesting failure of economic losers to halt progress in English economic history comes from the impact of the enclosure of common lands. Land has not always been privately owned as property. In much of Africa land is still owned communally, rather than individually, and this was true in Medieval Britain. Starting around 1550 however an ‘enclosure movement’ gathered pace where ‘common land’ was divided between cultivators and privatized. By 1850 this process of enclosures had made practically all of Britain private property. Enclosure was a heterogenous process [Overton (1996, p. 147)] and it also took place at different times in different places. Nevertheless, most of it was in two waves, the so called ‘Tudor enclosures’ between 1550 and 1700 and the ‘parliamentary enclosures’ in the century after 1750. “From the mid-eighteenth century the most usual way in which common rights were removed was through a specific act of parliament for the enclosure of a particular locality. Such acts . . . made the process easier because enclosure could be secured provided the owners of a majority (four fifths) of the land, the lord of the manor, and the owner of the tithe agreed it should take place. Thus the law of
436
D. Acemoglu et al.
parliament (statue law) only took account of the wishes of those owning land as opposed to the common law which took account of all those who had both ownership rights and use rights to land. Moreover . . . in some parishes the . . . majority could be held by a single landowner . . . parliamentary enclosure often resulted in a minority of owners imposing their will on the majority of farmers” [Overton (1996, p. 158), italics in original]. The historical evidence is unanimous that the incentive to enclose was because “enclosed land was worth more than open common field land . . . the general consensus has been that rents doubled” [Overton (1996, p. 162)]. More controversial is the source of this increase in rent. Overton continues (pp. 162–163) “The proportion of profits taken as rents from tenants by landlords is the outcome of a power struggle between the two groups, and the increase in rent with enclosure may simply reflect an increase in landlord power”. Allen (1982, 1992) showed, in his seminal study of the enclosure movement in the South Midlands, that the main impact was a large increase in agricultural rents and a redistribution of income away from those cultivators who had previously used the commons. The enclosure of common land thus led to a huge increase in inequality in early modern England. Many peasants and rural dwellers had their traditional property rights expropriated. In protest, groups of citizens dispossessed by enclosure attempted to oppose it through collective action and riots – attempting to influence the exercise of political power. These groups were no match for the British state, however. Kett’s rebellion of 1549, the Oxfordshire rebellion of 1596, the Midland Revolt of 1607, and others up to the Swing Riots of 1830–1831 were all defeated [see Charlesworth (1983)]. The presence of economic losers did not prevent this huge change in economic institutions and income distribution. 6.4. The inseparability of efficiency and distribution Commitment problems in the use and the allocation of political power therefore introduce a basic trade-off between efficiency and distribution. For example, when lack of commitment causes hold-ups, those who hold political power know that people will not have the right incentives to invest so growth will be low. In response to this, they might voluntarily give away their power or try to create political institutions that restricted their power. Such a change in political institutions would create better investment incentives. Though this situation is hypothetically possible and has formed the basis for some theories of institutional change [e.g., Barzel (2001)] it appears to be insignificant in reality. Even faced with severe underinvestment, political elites are reluctant to give away their power because of its distributional implications, i.e., because this would reduce their ability to extract rents from the rest of society. Thus poor economic institutions, here lack of property rights and hold-up, persist in equilibrium because to solve the problem, holders of political power have to voluntarily constrain their power or give it away. This may increase the security of property in society and increase incentives to
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
437
invest, but it also undermines the ability of rulers to extract rents. They may be better off with a large slice of a small pie. Similar phenomena are at work when there are either political or economic losers. In the first case, namely a situation where political power holders anticipate being political losers, promoting good institutions directly reduces the political power and rents of incumbents and a similar trade-off emerges. Adopting efficient economic institutions will stimulate growth, but when the political status quo is simultaneously eroded the amount of rent accruing to the initially powerful may fall. In the second case, the incomes of those with political power to determine economic institutions falls directly when better economic institutions are introduced. In the absence of credible commitments to sidepayments, those whose incomes fall when better economic institutions are introduced have an incentive to block such institutions. Because commitment problems seem so endemic in collective choice and politics, it seems natural to believe that institutional change has significant distributional consequences and as a result there will be conflict over the set of institutions in society. 6.5. Comparative statics Our analysis so far has made some progress towards our theory of differences in economic institutions. Although our full theory is yet to be developed in the later sections, the different mechanisms discussed in this section already point out the major comparative static implications of our approach regarding when economic institutions protecting the property rights of a broad cross-section of society are likely to be adopted, and when they are likely to be opposed and blocked. We now briefly discuss these comparative statics. Hold-up, political loser and economic loser considerations lead to some interesting comparative static results which can be derived by considering the political institutions that lie behind these phenomena. 1. First, the perspective of hold-ups immediately suggests that situations in which there are constraints on the use of political power, for example, because there is a balance of political power in society or a form of separation of powers between different power-holders, are more likely to engender an environment protecting the property rights of a broad cross-section of society. When political elites cannot use their political power to expropriate the incomes and assets of others, even groups outside the elite may have relatively secure property rights. Therefore, constraints and checks on the use of political power by the elite are typically conducive to the emergence of better economic institutions 2. Second, a similar reasoning implies that economic institutions protecting the rights of a broad cross-section are more likely to arise when political power is in the hands of a relatively broad group containing those with access to the most important investment opportunities. When groups holding political power are narrower, they may protect their own property rights, and this might encourage their own
438
D. Acemoglu et al.
investments, but the groups outside the political elites are less likely to receive adequate protection for their investments [see Acemoglu (2003b)]. 3. Third, good economic institutions are more likely to arise and persist when there are only limited rents that power holders can extract from the rest of society, since such rents would encourage them to opt for a set of economic institutions that make the expropriation of others possible. 4. Finally, considerations related to issues of political losers suggest that institutional reforms that do not threaten the power of incumbents are more likely to succeed. Therefore, institutional changes that do not strengthen strong opposition groups or destabilize the political situation are more likely to be adopted. 6.6. The colonial experience in light of the comparative statics We now briefly return to the colonial experience, and discuss how the comparative statics discussed here shed light on the differences in economic institutions across the former colonies and the institutional reversal. The second comparative static result above suggests a reason why better economic institutions developed in places where Europeans settled. In these societies, a relatively broad-based group of Europeans came to dominate political power, and they opted for a set of economic institutions protecting their own property rights. In contrast, in places where Europeans did not settle, especially where they were a small minority relative to a large indigenous population, they did not have the incentives to develop good economic institutions because such institutions would have made it considerably more difficult for them to extract resources from the rest of society. The third comparative static suggests an important reason why in places with more wealth, resources and also a high density of indigenous population to be exploited, Europeans were more likely to opt for worse institutions, without any protection for the majority of the population, again because such institutions facilitated the extraction of resources by the Europeans. The first comparative static result, in turn, is related to the persistence of the different types of economic institutions that Europeans established, or maintained, in different colonies. In colonies where Europeans settled in large numbers, they also developed political institutions placing effective checks on economic and political elites. In contrast, the political institutions in colonies with high population density, extractive systems of production, and few Europeans, concentrated power in the hands of the elite, and built a state apparatus designed to use coercion against the majority of the population. These different political institutions naturally implied different constraints on political and economic elites. In the former set of colonies, there were constraints on the development of economic institutions that would favor a few at the expense of the majority. Such constraints were entirely absent in the latter set of colonies. Finally, the fourth comparative static is useful in thinking about why many colonies did not attempt to change their economic institutions during the nineteenth century when new economic opportunities made their previous system based on forced labor, slavery,
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
439
or tribute-taking much less beneficial relative to one encouraging investment in industry and commerce. Part of the answer appears to lie in the fact that the political power of the elites, for example of the plantation owners in the Caribbean, was intimately linked to the existing economic system. A change in the economic system would turn them into political losers, an outcome they very much wanted to avoid. 6.7. Reassessment of the social conflict view So far we have shown that the econometric evidence is convincing that differences in economic institutions are the root cause of differences in prosperity. We then argued that although there are different approaches which can account for variation in economic institutions, the most plausible approach is the social conflict view. Though we believe that there certainly are instances where history and ideology matter for the institutional structure of society, and clearly institutions are highly persistent, the most promising approach to understanding why different countries have different institutions is to focus on choices and their subsequent consequences. The social conflict view emphasizes the distributional implication of economic institutions and how commitment problems imply that efficiency and distribution cannot be separated. Hence the fundamental conflict within society over the nature of economic institutions has important implications for economic performance. Some economic institutions will promote growth, but they will not necessarily benefit all groups in society. Alternative economic institutions may induce economic stagnation, but may nevertheless enrich some groups. Which set of institutions results and whether or not a society prospers will be determined by which of these groups has the political power to get the institutions that differentially benefit them. At this point we have therefore substantiated the first three points we made in the introduction. To develop our theory of economic institutions further we need to be more specific about political power – where it comes from and why some people have it and not others. We undertake this task in Section 8. Before doing this however the next section discusses three important historical examples of the evolution of economic institutions. We use these examples to show the explanatory power of the social conflict view and to begin to illustrate in concrete settings how political power works.
7. The social conflict view in action We now discuss three important examples to bring out the fact that conflict over economic institutions is critical to the functioning of the economy and that this conflict stems, not from differences in beliefs, ideology or historical accidents, but from the impact of economic institutions on distribution. The examples also show that those with political power have a disproportionate effect on economic institutions and they illustrate how the distribution of political power is influenced by different factors. These factors include the allocation of de jure political power through the structure of political institutions and the ability of groups to solve the collective action problem, or exercise
440
D. Acemoglu et al.
what we called de facto political power. With these examples in mind in Section 8 we move to discuss in more detail the nature and sources of political power. 7.1. Labor markets A market – an opportunity for individuals to exchange a commodity or service – is obviously a fundamental economic institution relevant for development. As Adam Smith (1776) argued, markets allow individuals to take advantage of the benefits of specialization and the division of labor, and scholars such as Pirenne (1937) and Hicks (1969) argued that the expansion of markets was perhaps the driving forces in long-run development. In the history of Europe a key transformation was from feudal labor market institutions towards modern notions of a free labor market where individuals were able to decide who to work for and where to live. This process of institutional change was intimately connected to the transition from a whole set of feudal economic institutions to the economic institutions we think of as ‘capitalist’. Most historians see this as key to the economic take-off that began in the nineteenth century. It was the countries which had made the transition away from feudalism most completely, such as England, the Netherlands and France, thanks to the revolution of 1789, which developed most rapidly. It was those where feudalism was still in operation, such as Russia and Austria-Hungary, which lagged far behind. What can account for this differential evolution of feudalism? Scholars beginning with Postan (1937) saw the demographic collapse caused by the black death in the 1340’s as demolishing feudalism in Western Europe. By dramatically altering the land/labor ratio as approximately 40% of the population of Europe died [e.g., Cantor (2001)], the Black Death greatly increased the bargaining power of peasants and allowed them to negotiate a free status ending feudal obligations, particularly with respect to labor. Therefore, Postan’s demographic theory implicitly emphasizes the role of political power in the decline of feudalism: this set of economic institutions started to disappear when the political power of the peasants increased and that of lords declined. In fact, the distribution of power may be even more important in the whole story than Postan’s theory suggests. As first pointed out by Brenner (1976), the demographic theory of the decline feudalism is not consistent with the comparative evidence. Although demographic trends were similar all over Europe and “it is true that . . . in most of Western Europe serfdom was dead by the early sixteenth century. On the other hand, in Eastern Europe, in particular Pomerania, Brandenburg, East Prussia and Poland, decline in population from the late fourteenth century was accompanied by an ultimately successful movement towards imposing extra-economic controls, that is serfdom, over what had been, until then, one of Europe’s freest peasantries. By 1500 the same Europe-wide trends had gone a long way towards establishing one of the great divides in European history, the emergence of an almost totally free peasant population in Western Europe, the
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
441
debasement of the peasantry to unfreedom in Eastern Europe.” [Brenner (1976, p. 41)]. What can explain these divergent outcomes? Brenner notes (p. 51): “It was the logic of the peasant to try to use his apparently improved bargaining position to get his freedom. It was the logic of the landlord to protect his position by reducing the peasants’ freedom”. The outcome “obviously came down to a question of power” (p. 51); whether the peasants or the lords had more political power determined whether serfdom declined or became stronger. Although we are far from an understanding of the determinants of the relative structure of political power in different parts of Europe, Brenner suggests that an important element was the “patterns of the development of the contending agrarian classes and their relative strength in the different European societies: their relative levels of internal solidarity, their self-consciousness and organization, and their general political resources – especially their relationships to the non-agricultural classes (in particular, potential urban class allies) and to the state” (p. 52). To substantiate this view, Brenner studies how villages tended to be organized differently in Eastern Europe, there was “more of a tendency to individualistic farming; less developed organization of collaborative agricultural practices at the level of the village or between villages; and little of the tradition of the ‘struggle for commons rights’ against the lords which was so characteristic of western development” (p. 57). This differential organization was due to the process of initial occupation of these Eastern lands. Although many parts of Brenner’s analysis remain controversial, there is general agreement that the decline of feudalism and the transformation of European labor markets were intimately related to the political power of the key groups with opposing interests, the peasants and the lords [see, for example, Aston and Philpin (1985) on reactions to Brenner’s interpretation]. Feudal institutions, by restricting labor mobility and by removing the role of the labor market in allocating labor to jobs, undermined incentives and resulted in underdevelopment. But these same economic institutions created large rents for the aristocracy. As a consequence, aristocracies all over Europe attempted to maintain them. It was when their political power weakened that the process of transformation got underway. 7.2. Financial markets Much recent work on growth and development has focused on capital markets. Growth requires investment, so poor agents without access to financial markets will not have the resources to invest. Empirically many scholars have found correlations between the depth of financial markets and growth [see Levine (2005)] and absence of financial markets is at the heart of ambitious theories of comparative development by Banerjee and Newman (1993) and Galor and Zeira (1993). If the stress on financial markets and financial intermediation is correct, a central issue is to understand why financial systems differ. For example, studies of the development
442
D. Acemoglu et al.
of banking in the United States in the nineteenth century demonstrate a rapid expansion of financial intermediation which most scholars see as a crucial facilitator of the rapid growth and industrialization that the economy experienced. In his recent study Haber (2001, p. 9) found that in the United States, “In 1818 there were 338 banks in operation, with a total capital of $160 million-roughly three times as many banks and bank capital as in 1810. Circa 1860, the United States had 1,579 banks, with a total capital of $422.5 million. Circa 1914 there were 27,864 banks in the United States. Total bank assets totaled $27.3 billion”. One might see this rapid expansion of banking and financial services as a natural feature. Yet Haber (2001) shows that the situation was very different in Mexico (p. 24). “Mexico had a series of segmented monopolies that were awarded to a group of insiders. The outcome, circa 1910 could not have been more different: the United States had roughly 25,000 banks and a highly competitive market structure; Mexico had 42 banks, two of which controlled 60 percent of total banking assets, and virtually none of which actually competed with another bank.” The explanation for this huge difference is not obvious. The relevant technology was certainly readily available everywhere and it is difficult to see why the various types of moral hazards or adverse selection issues connected with financial intermediation should have limited the expansion of banks in Mexico but not the United States. Haber then shows that (p. 9), “at the time that the U.S. Constitution was put into effect in 1789, . . . [U.S. banking] was characterized by a series of segmented monopolies that shared rents with state governments via taxes or state ownership of bank stock. In some cases, banks also shared rents directly with the legislators who regulated them.” This structure, which looked remarkably like that which arose subsequently in Mexico, emerged because state governments had been stripped of revenues by the Constitution. In response, states started banks as a way to generate tax revenues. State governments restricted entry “in order to maximize the amount of rent earned by banks, rent which would then be shared with the state government in the form of dividends, stock distributions, or taxes of various types”. Thus in the early nineteenth century, U.S. banks evolved as monopolies with regulations aimed at maximizing revenues for the state governments. Yet this system did not last because states began competing among themselves for investment and migrants. “The pressure to hold population and business in the state was reinforced by a second, related, factor: the broadening of the suffrage. By the 1840s, most states had dropped all property and literacy requirements, and by 1850 virtually all states (with some minor exceptions) had done so. The broadening of the suffrage, however, served to undermine the political coalitions that supported restrictions on the number of bank charters. That is, it created a second source of political competition-competition within states over who would hold office and the policies they would enact.”
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
443
The situation was very different in Mexico. After 50 years of endemic political instability the country unified under the highly centralized 40 year dictatorship of Porfirio Diaz until the Revolution in 1910. In Haber’s argument political institutions in the United States allocated political power to people who wanted access to credit and loans. As a result they forced state governments to allow free competitive entry into banking. In Mexico political institutions were very different. There were no competing federal states and the suffrage was highly restrictive. As a result the central government granted monopoly rights to banks who restricted credit to maximize profits. The granting of monopolies turned out to be a rational way for the government to raise revenue and redistribute rents to political supporters [see North (1981, Chapter 3)]. A priori, it is possible that the sort of market regulation Haber found in Mexico might have been socially desirable. Markets never function in a vacuum, but rather within sets of rules and regulations which help them to function. Yet it is hard to believe that this argument applies to Mexico [see also Maurer (2002)]. Haber (2001) documents that market regulation was aimed not at solving market failures and it is precisely during this period that the huge economic gap between the United States and Mexico opened up [on which see Coatsworth (1993), Engerman and Sokoloff (1997)]. Indeed, Haber and Maurer (2004) examined in detail how the structure of banking influenced the Mexican textile industry between 1880 and 1913. They showed that only firms with personal contacts with banks were able to get loans. They conclude (p. 5): “Our analysis demonstrates that textile mills that were related to banks were less profitable and less technically efficient than their competitors. Nevertheless, access to bank credit allowed them to grow faster, become larger, and survive longer than their more productive competitors. The implication for growth is clear: relatively productive firms lost market share to relatively unproductive (but bank-related) competitors.” Despite the fact that economic efficiency was hurt by regulations, those with the political power were able to sustain these regulations. 7.3. Regulation of prices As our final example we turn to the regulation of prices in agricultural markets (which is intimately related to the set of agricultural policies adopted by governments). The seminal study of agricultural price regulation in Africa and Latin America is by Bates (1981, 1989, 1997). Bates (1981) demonstrated that poor agricultural performance in Ghana, Nigeria and Zambia was due to government controlled marketing boards systematically paying farmers prices for their crops much below world levels. “Most African states possess publicly sanctioned monopsonies for the purchase and export of agricultural goods . . . These agencies, bequeathed to the governments of the independent states by their colonial predecessors, purchase cash crops
444
D. Acemoglu et al.
for export at administratively determined domestic prices, and then sell them at the prevailing world market prices. By using their market power to keep the price paid to the farmer below the price set by the world market, they accumulate funds from the agricultural sector” [Bates (1981, p. 12)]. The marketing boards made surpluses which were given to the government as a form of taxation. Bates (1981, p. 15) notes “A major test of the intentions of the newly independent governments occurred . . . [when] between 1959–60 and 1961–62, the world price of cocoa fell approximately £50 a ton. If the resources generated by the marketing agencies were to be used to stabilize prices, then surely this was the time to use the funds for that purpose. Instead . . . the governments of both Ghana and Nigeria passed on the full burden of the drop in price to the producers.” Bates continues “Using the price setting power of the monopsonistic marketing agencies, the states have therefore made the producers of cash crops a significant part of their tax base, and have taken resources from them without compensation in the form of interest payments or of goods and services returned” (pp. 181–189). As a result of this pernicious taxation, reaching up to 70% of the value of the crop in Ghana in the 1970s, investment in agriculture collapsed as did output of cocoa and other crops. In poor countries with comparative advantage in agriculture such a situation mapped into negative rates of economic growth. Why were resources extracted in this way? Though part of the motivation was to promote industrialization, the main one is to generate resources that could be either expropriated or redistributed to maintain power “governments face a dilemma: urban unrest, which they cannot successfully eradicate through co-optation or repression, poses a serious challenge to their interests . . . Their response has been to try to appease urban interests not by offering higher money wages but by advocating policies aimed at reducing the cost of living, and in particular the cost of food. Agricultural policy thus becomes a byproduct of political relations between governments and urban constituents.” [Bates (1981, p. 33)]. In contrast to the situation in Ghana, Zambia and Nigeria, Bates (1981, 1989, 1997) showed that agricultural policy in Kenya and Colombia over this period was much more pro-farmer. The difference was due to who controlled the marketing board. In Kenya, farmers were not smallholders, as they were in Ghana, Nigeria and Zambia, and concentrated landownership made it much easier to solve the collective action problem. Moreover, farming was important in the Kikuyu areas, an ethnic group closely related to the ruling political party, KANU, under Jomo Kenyatta [Bates (1981, p. 122)]. Farmers in Kenya therefore formed a powerful lobby and were able to guarantee themselves high prices. Even though the government of Kenya engaged in land reform after independence
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
445
“80% of the former white highlands were left intact and . . . the government took elaborate measures to preserve the integrity of the large-scale farms . . . [which] readily combine in defense of their interests. One of the most important collective efforts is the Kenya National Farmer’s Union (KNFU) . . . The organization . . . is dominated by the large-scale farmers . . . [but] it can be argued that the KNFU helps to create a framework of public policies that provides an economic environment favorable to all farmers.” [Bates (1981, pp. 93–94)]. Bates concludes (p. 95) that in Kenya “large farmers . . . have secured public policies that are highly favorable by comparison to those in other nations. Elsewhere the agrarian sector is better blessed by the relative absence of inequality. But is also deprived of the collective benefits which inequality, ironically, can bring.” In Colombia, farmers were favored because of competition for their votes from the two main political parties. Bates (1997, p. 54) notes “Being numerous and small, Colombia’s coffee producers, like peasants elsewhere, encountered formidable costs of collective action. In most similar instances such difficulties have rendered smallholders politically powerless. And yet . . . Colombia’s peasants elicited favorable policies from politicians, who at key moments themselves bore the costs of collective action, provisioning the coffee sector with economic institutions and delegating public power to coffee interests.” How could the coffee growers gain such leverage over national policy? “A major reason they could do so . . . is because the structure of political institutions, and in particular the structure of party competition, rendered them pivotal, giving them the power over the political fortunes of those with ambition for office and enabling them to make or break governments. They thereby gained the power to defeat government officials who sought to orchestrate or constrain their behavior.” [Bates (1997, pp. 51, 54)]. A telling piece of evidence in favor of this thesis is that during the 1950s when a civil war broke out between the two parties, there was five years of military rule and policy turned decisively again the coffee growers, only to switch back again with the peaceful resumption of democracy in 1958. 7.4. Political power and economic institutions These three examples of the creation of economic institutions have certain features in common. All these institutions, labor market regulation/feudalism, the rules governing financial market development, and agricultural price regulation, clearly reflect the outcome of conscious choices. Feudalism did not end in England for incidental or ideological reasons, but because those who were controlled and impoverished by feudal
446
D. Acemoglu et al.
regulations struggled to abolish them. In Eastern Europe the same struggle took place but with a different outcome. Similarly, Mexico did not end up with different financial institutions than the United States by accident, because of different beliefs about what an efficient banking system looked like, or because of some historical factor independent of the outcome. The same is true for differences in economic policies in Kenya and Ghana. Moreover, different sets of economic institutions arising in different places cannot be argued to be efficient adaptations to different environments. Most historians believe that the persistence of feudal institutions in Eastern Europe well into the nineteenth century explains why it lagged far behind Western Europe in economic development. The difference between the financial institutions of Mexico and the United States also plausibly played a role in explaining why they diverged economically in the nineteenth century. The same holds with respect to agricultural price regulation. The driving force behind all three examples is that economic institutions are chosen for their distributional consequences. Which specific economic institutions emerge depends on who is able to get their way – who has political power. In England, peasant communities had developed relatively strong local political institutions and were able to consolidate on the shock of the Black Death to put an end to feudal regulations. In Eastern Europe it was the lords who had relatively more power and they were able to intensify feudalism in the face of the same demographic shock [as Domar (1970) pointed out, the Black Death actually made serfdom more attractive to the lords even if at the same time it increased the bargaining power of the peasants]. In the case of banking in the nineteenth century, Haber’s research shows while the authoritarian regime in Mexico had the political power to freely create monopolies and create rents in the banking industry, the United States was different because it was federal and much more democratic. The political institutions of the United States prevented politicians from appropriating the rents that could flow from the creation of monopolies. Finally, in Bates’s analysis, distortionary price regulations arose in Ghana and Zambia, but not in Kenya and Colombia, because in the latter countries agricultural producers had more political power and so could prevent the distortionary policies that would harm their interests. It is also useful to consider in the context of these examples the mechanisms we discussed in Section 6 which underlie the adoption of inefficient economic institutions. Why could not the peasants and lords of feudal Europe negotiate and allow the introduction of a set of economic institutions that would have given peasants incentives to innovate and would have allowed for the efficient allocation of labor? Why could not either the lords have promised not to expropriate any benefits that accrued from innovation, or alternatively the peasants agreed to compensate the lords if feudal labor institutions were abolished? Though it is difficult to find direct evidence on such counterfactuals from the Medieval period, the most plausible explanation is that such ‘deals’ were impossible to make credible. The political power of the lords was intimately connected to feudal institutions and thus dismantling these would not only have increased peasant incentives to innovate, but would also have dramatically altered the balance of political power and the distribution of rents in society. Moreover, under feudal regulations peasants were tied to the land. The introduction of free labor mobility would have
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
447
given workers an exit option, thus increasing their bargaining power with the lords over the division of output. Thus lords might anticipate being both political and economic losers from the ending of feudalism, even if total output would have increased. In the case of agricultural price regulation, similar arguments are plausible. Cocoa farmers in Ghana would not have believed promises by governments that they would not expropriate the fruits of higher investment, and the governments themselves would not have believed promises by the farmers to compensate them if they left office. Moreover, efficient sets of economic institutions in Ghana or Nigeria would have strengthened the economic base of the rural sector at the expense of the political power of the then dominant urban sector. Indeed, for Ghana in the 1960s, we have direct evidence from the urban economy that the threat of being a political loser led to inefficient economic institutions. This emerges in the analysis of Killick (1978, p. 37) of the attempt by the government of Kwame Nkrumah to promote industrialization. Killick notes: “Even had there been the possibility [of creating an indigenous entrepreneurial class] it is doubtful that Nkrumah would have wanted to create such a class, for reasons of ideology and political power. He was very explicit about this saying ‘we would be hampering our advance to socialism if we were to encourage the growth of Ghanian private capitalism in our midst’. There is evidence that he also feared the threat that a wealthy class of Ghanaian businessmen might pose to his own political power.” Further evidence on the importance of political loser considerations comes from E. Ayeh-Kumi one of Nkrumah’s main economic advisers who noted after the coup that ousted Nkrumah in 1966 that Nkrumah: “informed me that if he permitted African business to grow, it will grow to becoming a rival power to his and the party’s prestige, and he would do everything to stop it, which he actually did” [Killick (1978, p. 60)]. In this context, it is interesting that Nkrumah’s solution to consolidate his power was to limit the size of businesses that Ghanaians could own. This caused problems for his industrialization policy which he got round by allowing foreign businessmen to enter Ghana. Though this was inconsistent with his aggressively nationalistic and antiimperialistic rhetoric, these businessmen did not pose a domestic political threat. Killick notes “Given Nkrumah’s desire to keep Ghanaian private businesses small, his argument that ‘Capital investment must be sought from abroad since there is no bourgeois class amongst us to carry on the necessary investment’ was disingenuous” (p. 37). He goes on to add that Nkrumah “had no love of foreign capitalists but he preferred to encourage them rather than local entrepreneurs, whom he wished to restrict” (p. 40). All these examples show that the distribution of political power in society is crucial for explaining when economic institutions are good and when they are bad. But where does political power come from and who has political power? In addressing these questions we will develop our theory of economic institutions. In a theory based on social conflict where economic institutions are endogenous, it will be to differences in political institutions and the distribution of political power that we must look to explain variation in economic institutions.
448
D. Acemoglu et al.
8. A theory of institutions 8.1. Sources of political power Who has political power and where does it come from? As we noted in the Introduction (Section 1.2, point 4), political power comes from two sources. First, an individual or group can be allocated de jure power by political institutions. But institutions are not the only source of power. A second type of political power accrues to individuals or groups if they can solve the collective action problem, create riots, revolts, or demonstrations, own guns, etc. We call this type of power de facto political power [see Acemoglu and Robinson(2003, Chapter 5)]. Actual political power is the composition, the joint outcome, of de jure and de facto power. To see how this works out in practice, consider the situation in Chile in the early 1970’s. Salvador Allende was elected President with a plurality of the popular vote. The formal political institutions of democracy in Chile allocated power to him to propose legislation, issue decrees, etc. Consequently, even though he did not have an absolute majority in congress, Allende had a great deal of de jure political power. Political power is not just de jure however; it does not simply stem from political institutions. Allende, despite being empowered under the Chilean Constitution, was overthrown by a military coup in 1973. Here, the military under the leadership of General Pinochet, were able to use brute force and guns to over-ride the formal political institutions. The ability to use force is one example of de facto political power. As we suggested in the introduction, the relationship between political power and economic and political institutions is complex and dynamic. Consider the example we discussed in Section 7.2, the research by Haber on the comparative financial evolution of Mexico and the United States in the nineteenth century. Haber traced the different evolution of economic institutions to differences in initial political institutions. These political institutions led to different distributions of power and this was critical for the emergence of good financial institutions in the United States, whereby those who benefited from a competitive banking industry were able to force politicians to provide the rules which would guarantee it. But where did these differences in political institutions come from? These differences were partly a result of political events in the nineteenth century, and partially a result of different colonial political institutions. In the United States, during the initial phase of colonization in the early seventeenth century very low population density and lack of easily exploitable resources forced colonizing companies and the British state to make both economic and political concessions; they granted the settlers access to land and accepted the formation of representative democratic institutions [Morgan (1975)]. Consequently, even at independence the United States had relatively democratic political institutions [Keyssar (2000)]. Moreover, the initial egalitarian distribution of assets and the high degree of social mobility made for a situation where, at least in the northern states, the distribution of economic resources, and thus de facto power, was relatively equal. The relatively representative political institutions therefore persisted and were supported by the balance of de facto power in society.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
449
In Mexico there were very different initial conditions during the colonial period with a large indigenous population and rich silver mines to exploit. This led to a much more hierarchical and authoritarian balance of political power and very different colonial economic institutions [see Engerman and Sokoloff (1997)]. These conditions fed into the different institutional structures at independence, the United States with its constitution, checks and balances and federalism, Mexico with its much more centralized, unchecked, unbalanced and absolutist state. These different political institutions then led to very different economic institutions and economic outcomes after independence. Thus, in some ultimate sense, the source of different political institutions were different initial conditions during the colonial period. Consider now the evidence presented by Bates. Agricultural policies were better in Kenya because large farmers could solve the collective action problem and exercise de facto political power. But the main reason for the existence of large farms was that British settlers expropriated the land from Africans during the expansion of colonialism [see Berman and Lonsdale (1992)]. Thus previous combinations of formal political institutions (colonial institutions) and de facto power (the military might of the British Empire) determined economic institutions, feeding into the future distribution of de facto power even after the nature of de jure power changed dramatically with independence. We can now see that these examples substantiate the dynamic model that we sketched in Section 1.2. There we showed that at any date, political power is shaped by political institutions, which determine de jure power, and the inherited distribution of resources, which affect the balance of de facto power. Political power then determines economic institutions and economic performance. It also influences the future evolution of political power and prosperity. Economic institutions determine the distribution of resources at that point, which, in turn, influences the distribution of de facto power in the future. Similarly, the distribution of power at any point determines not just the economic institutions then, but also the future political institutions. Thus the allocation of political power at one date, because of the way it influences the distribution of resources and future political institutions, has a crucial effect on the future allocation of both de facto and de jure political power. Both the comparison Haber made between Mexico and the United States, and that which Bates made between Ghana, Zambia, Kenya and Colombia illustrate this diagram in action. They show how political institutions and de facto power combine to generate different set of economic institutions, how these institutions determine both the distribution of resources and the growth rate of the economy, and how power and institutions evolve over time, often in ways that tend to reinforce particular initial conditions. 8.2. Political power and political institutions The examples we discussed above showed how political power depends on political institutions and de facto power, and how this determines economic institutions. Moreover, we saw that at any time the pre-existing economic institutions will be an important de-
450
D. Acemoglu et al.
terminant of the distribution of de facto power. The final element to emphasize is how political institutions evolve over time and how they influence the distribution of political power. To see why political institutions are so important as a source of political power think of a situation where a group, say the Chilean army in the early 1970s, has a great deal of de facto power. Indeed, it has so much de facto power that it can overrule the Chilean Constitution, making the political institutions largely irrelevant. In fact in Chile the de facto power of the military was able to overthrow the legitimate government and completely reverse the economic policies and economic institutions chosen by the Allende government (including land reform and mass nationalization of industry). Not only did the military reverse the economic institutions preferred by Allende and the groups who elected him, they then implemented their own preferred set of economic institutions, in particularly deregulating the trade regime and the economy. Yet the Pinochet regime was heavily concerned with political institutions, and in 1980 Pinochet re-wrote the constitution. If de facto power was decisive in Chile what is the role for political institutions? If the constitution can be overthrown, why bother to re-write it? The secret to this lies in the intrinsically transitory nature of de facto power.13 Yes, the military were able to organize a coup in 1973 but this was only because times were uniquely propitious. There was a world-wide economic crisis, and factions of the military that opposed the coup could be marginalized. Moreover, the United States government at the time was happy to encourage and endorse the overthrow of a socialist government, even if it had been democratically elected. The coming together of such circumstances could not be expected to happen continually, hence once Chilean society re-democratized, as it did after 1990, the military would not be able to continually threaten a coup. In response to this Pinochet changed the political institutions in order to attempt to lock in the power of the military, and thus the economic institutions that he/they preferred. Therefore, the important role for political institutions is that they influence the future allocation of political power. This dynamic role is crucial because it explains the major desire of agents to change political institutions when they get the chance – this is how they can attempt to enduringly alter the balance of political power in their favor [see Acemoglu and Robinson (2003)].
13 The empirical literature on the collective action problem has recognized that the difficulty of solving
the collective action problems lead collective action to typically be transitory. Lichbach (1995, p. 17) notes “collective action, if undertaken on a short-term basis, may indeed occur; collective action that requires long periods to time does not . . . Given that most people’s commitments to particular causes face inevitable decline, most dissident groups are ephemeral, most dissident campaigns brief”. This transitory nature of collective action is echoed by Tarrow (1991, p. 15) who notes “the exhaustion of mass political involvement”, while Ross and Gurr (1989, p. 414) discuss political “burnout”. Similarly, Hardin (1995, p. 18) argues that “the extensive political participation of civil society receives enthusiastic expression only in moments of state collapse or great crisis. It cannot be maintained at a perpetually high level.”
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
451
8.3. A theory of political institutions We now have in place the outlines of our theory of institutions. There are seven points to emphasize, paralleling the discussion in Section 1.2 and our diagrammatic exposition there. First, individuals have preferences over economic institutions because of the allocation of resources that these institutions induce. Second, people’s preferences typically do not agree because efficiency and distribution cannot be separated. Different economic institutions will benefit different groups, and this will determine the preferences of these individuals and groups with respect to economic institutions. Third, the problem of commitment explains why efficiency and distribution are inseparable. Economic institutions are collective choices, and they are chosen and sustained by the state. Since there is no third party to enforce the decisions of the state, problems of commitment are particularly severe in the political realm. Fourth, the equilibrium structure of economic institutions will therefore be determined by who has the power to get their way, i.e., who can create and sustain economic institutions that benefit themselves. The distribution of political power thus determines economic institutions, the allocation of resources and the rate of economic growth. Fifth, political power has two forms: de jure power determined by the political institutions, such as the constitution and the electoral rules, and de facto power, which stems from the ability to solve the collective action problem, mobilize weapons, etc. De facto power can influence political outcomes independently of the political institutions, and its distribution often critically determines how a given set of institutions works in practice and whether or not they are actually obeyed. Sixth, the distribution of de facto political power at any date is influenced to a large degree by the distribution of resources in society, since those with greater resources can command more power both through legitimate and intimate means, and perhaps can also solve the collective action problem more efficiently. Naturally, the distribution of resources at this point is influenced by economic institutions and economic outcomes in the past. Finally, political institutions are also endogenous; the current balance of political power, incorporating both de jure and de facto elements, also determines future political institutions. Political institutions are important because they allocate, at least within the limits defined by the exercise of future de facto power, the allocation of future de jure political power. Since de facto power, because of the nature of the collective action problem, is intrinsically transitory and difficult to wield, political institutions are often crucial in creating a source of durable political power. This makes it very attractive for groups to use their de facto political power to change political institutions so as to modify the distribution of future political power in their favor.
452
D. Acemoglu et al.
9. The theory in action We now consider two more examples that further demonstrate our theory of institutions in action. Like the examples discussed in Section 7, these examples contain all the elements of our theory laid out in a skeletal way in Section 1.2. They show the role of political power in determining economic institutions, they demonstrate the different factors, both de facto and de jure, that determine political power, and they illustrate how de facto political power is often used to change political institutions in order to influenced the future distribution of de jure political power. 9.1. Rise of constitutional monarchy and economic growth in early modern Europe Our first example is the rise of constitutional monarchy in Europe. In the medieval period most European nations were governed by hereditary monarchies. However, as the feudal world changed, various groups struggled to gain political rights and reduce the autocratic powers of monarchies. In England, this process began as early as 1215 when King John was forced by his barons to sign the Magna Carta, a document which increased the powers of the barons, introduced the concept of equality before the law, and forced subsequent kings to consult with them. Many other European nations also developed ‘parliaments’ which kings could summon to discuss taxation or warfare [see Graves (2001), Ertman (1997)]. Nevertheless, the movement towards limited, constitutional monarchy was not linear or simple. Indeed, in France, certainly from the beginning of Louis XIV’s reign in 1638, a more powerful absolutist monarchy appeared with very few controls. Indeed the feudal French parliaments, the Estates General, were not summoned between 1614 and 1788, just before the Revolution. In England, the Tudor monarchs, particularly Henry VIII and then Elizabeth I, followed by the first Stuart kings, James I and Charles I, also attempted to build an absolutist monarchy. They failed, however, mostly because of Parliament, which blocked attempts to concentrate power. The constitutional outcome in England was settled by the Civil War from 1642–1651 and the Glorious Revolution in 1688. In the first of these conflicts the forces of Parliament defeated those loyal to Charles I and the king was beheaded. In 1660 the monarchy was restored when Charles II became king, but his brother James II was deposed in 1688 and Parliament invited William of Orange to become king. Other places in Europe, particularly the Netherlands, saw similar developments to those in England. Under the Dukes of Burgundy, the Netherlands had won a considerable amount of political and economic freedom, particularly under the Grand Privilege of 1477 which gave the States General of the Burgundian Netherlands the right to gather on their own initiative and curbed the right of the ruler to raise taxes. However, the Netherlands were inherited by the Hapsburgs through marriage, and by 1493 Maximilian of Hapsburg had reversed the Grand Privilege. After 1552, war with France increased the Hapsburgs’ fiscal needs and led them to impose a large tax burden on the Netherlands, already a prosperous agricultural and mercantile area. Growing fiscal and
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
453
religious resentment in 1572 led to a series of uprisings against the Hapsburgs, mostly orchestrated by commercial interests. These culminated in the War of Independence which was finally won in 1648. While England and the Netherlands were developing limited constitutional governments, Spain and Portugal were moving in the same direction as France, towards greater absolutism. Davis (1973a, p. 66) notes [in Castille] “the king ruled subject only to weak constitutional restraints. In the first decades of the sixteenth century the crown had reduced the pretensions of the Castillian nobility and towns, so that the representative body, the Cortes, could obstruct but not in the last resort prevent royal tax raising.” These differential institutional trajectories were of enormous consequence. The economies of the Netherlands and England moved ahead of the rest of Europe precisely because these countries developed limited, constitutional government. This form of government led to secure property rights, a favorable investment climate and had rapid multiplier effects on other economic institutions, particularly financial markets [see, e.g., North and Weingast (1989), de Vries and van der Woude (1997)]. While the Netherlands and Britain prospered, France was convulsed by the French Revolution, and by the nineteenth century Spain and Portugal were impoverished backward nations. How can we account for these diverging paths in the early modern period? Why did England and the Netherlands develop limited constitutional rule, while France, Spain and Portugal did not? We proposed an explanation in Acemoglu, Johnson and Robinson (2005) related to the differential responses of these countries to the opportunities of ‘Atlantic trade’, that is, overseas trade and colonial activity unleashed by the discovery of the New World and the rounding of the Cape of Good Hope at the end of the fifteenth century. All five nations engaged in Atlantic trade, but they did so in different ways, with very different implications for the organization of society, political institutions and subsequent economic growth. In England “most trade was carried on by individuals and small partnerships, and not by the Company of Merchant Adventurers, the Levant Company . . . or others of their kind” [Davis (1973b, p. 41)]. At least by 1600 there was quite free entry into the English merchant class. The same was true in the Netherlands. In contrast, Cameron (1993, p. 127) describes the Portuguese situation as follows: “The spice trade in the East Indies of the Portuguese Empire was a crown monopoly; the Portuguese navy doubled as a merchant fleet, and all spices had to be sold through the Casa da India (India House) in Lisbon . . . no commerce existed between Portugal and the East except that organized and controlled by the state”. In Spain, similarly, colonial trade was a monopoly of the Crown of Castille, which they delegated to the Casa de Contratación (House of Trade) in Seville. This merchants guild was closely monitored by the government [Parry (1966, Chapter 2)]. The main aim of these regulations was to make sure that all of the gold and silver from the Americas flowed back to Spain, creating a source of direct tax revenues for the crown. As a result, Latin American colonies were forbidden to buy manufactured goods from anywhere other than Spain, and all exports and imports had to pass through controlled channels. For example, until the Bourbon reforms of the mid eighteenth cen-
454
D. Acemoglu et al.
tury, nothing could be exported directly from Buenos Aires, and if somebody produced anything for export on the Pampas, it had to be carried over the Andes and exported from Lima in Peru! The source of the differences in the organization of trade, in turn, reflected the different political institutions of these countries. At the time, the granting of trade monopolies was a key fiscal instrument to raise revenues; the more powerful monarchs could increase their revenues by granting trade monopolies or by directly controlling overseas trade, while weaker monarchs could not. At the turn of the fifteenth century, the crown was much stronger in France, Spain and Portugal than in Britain and the Netherlands, and this was the most important factor in the differences in the organization of overseas trade. In fact, when both Tudor and Stuart monarchs attempted to create monopolies similar to those in Spain and Portugal, this was successfully blocked by the English Parliament [see, for example, Hill (1969)]. Consequently, as world trade expanded in the sixteenth and early seventeenth centuries, in England and the Netherlands it enriched merchants engaged in overseas trade, but in France, Spain and Portugal it enriched the crown and groups allied with it. In England and the Netherlands, but not in France, Spain and Portugal, a new class of merchants (and gentry in England) arose with interests directly opposed to those of the Stuarts and the Hapsburgs, and this group was to play a central part in subsequent political changes. In the case of the Netherlands, de Vries and van der Woude (1997) argue that “urban economic interests ultimately believed it advantageous to escape the Hapsburg imperial framework” (p. 369), and that it was “the traditional pillars of the maritime economy . . . that supported and strengthened the young Republic in its hour of need” (p. 366). Moreover, in the case of Amsterdam, “[Hapsburgs’] opponents included most of the city’s international merchants . . . In 1578 a new Amsterdam city council threw the city’s lot in with the Prince of Orange . . . among the merchants returning from . . . exile were [those whose families] and several generations of their descendants would long dominate the city” (1997, p. 365). The expansion of world trade enriched and expanded precisely those groups within Dutch society most opposed to Hapsburg rule. Israel (1995, pp. 241–242) writes: “From 1590, there was a dramatic improvement in the Republic’s economic circumstances. Commerce and shipping expanded enormously, as did the towns. As a result, the financial power of the states rapidly grew, and it was possible to improve the army vastly, both qualitatively, and quantitatively, within a short space of time. The army increased from 20,000 men in 1588 to 32,000 by 1595, and its artillery, methods of transportation, and training were transformed” [see also Israel (1989, Chapter 3)]. By 1629, the Dutch were able to field an army of 77,000 men, 50% larger than the Spanish army of Flanders [Israel (1995, p. 507)]. As a consequence of the Dutch revolt, the Netherlands developed a republican form of government closely attuned to mercantile interests. De Vries and van der Woude (1997, p. 587) describe the new political elite following the Dutch Revolt as: “6 to 8% of urban households with incomes in excess of 1,000 guilders per year. This was the grote burgerij from whom was drawn the political and commercial leadership of the country. Here we find, first and
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
455
foremost, the merchants”, and point out how merchants dominated the governments of Leiden, Rotterdam and the cities in two largest states, Zeeland and Holland. In England, the Civil War and Glorious Revolution coincided with the great expansion of English mercantile groups into the Atlantic. The East India Company was founded in 1600 as the culmination of a series of efforts to develop trade routes with Asia. The 1620s saw the great expansion of tobacco cultivation in Virginia and this was shortly followed by the development of the highly profitable English sugar colonies in the Caribbean. Finally, in the 1650s the English began to take over the Atlantic slave trade. Both the Civil War and the Glorious Revolution were at root battles over the rights and prerogatives of the monarchy. In both cases new merchant interests predominantly sided with those in the gentry demanding restrictions on the powers of the monarchy in order to protect their property and commerce. The majority of merchants trading with the Americas and in Asia supported Parliament during the Civil War. Brunton and Pennington (1954, p. 62) also note “in the country as a whole there was probably a preponderance of Parliamentarian feeling among merchants”. Detailed analyses of the initial members of the Long Parliament in 1640 show that a significant majority of merchants supported the Parliamentarian cause [see Brenner (1973, 1993), Keeler (1954) and Brunton and Pennington (1954)]. Members of the Commons from the City of London (the main center of mercantile activity), as well as many non-London commercial constituencies, such as Southampton, Newcastle and Liverpool, supported Parliament against the King. These men included both professional merchants and aristocrats who invested in colonizing the Americas. These new merchants also provided the financial support needed by Parliament in the difficult early days of the war. They became the customs farmers for the new regime and therefore advanced tens of thousands of pounds that were essential in building up the army [Brenner (1973, p. 82)]. Pincus (1998, 2001, 2002) further documents the critical role of mercantile interests in the Glorious Revolution. He concludes (2002, p. 34) “England’s merchant community actively supported William’s plan for invasion, and provided a key financial prop to the regime in the critical early months”. He notes that James II favored the East India Company and granted various monopoly privileges, alienating the merchant class. Thus, “no wonder the merchant community poured money into William of Orange’s coffers in 1688” [Pincus (2002, pp. 32–33)]. The changes in the distribution of political power, political institutions and thus economic institutions that took place in England and the Netherlands had no counterparts in countries with relatively absolutist institutions, like Spain and Portugal, where the crown was able to closely control the expansion of trade. In these countries it was the monarchy and groups allied with it that were the main beneficiaries of the early profits from Atlantic trade, and groups favoring political and economic change did not become strong enough to induce such change. As a result, only in the Netherlands and England did constitutional rule emerge, and only in these two countries were property rights secure. As a result it was these same two countries that prospered.
456
D. Acemoglu et al.
Why could the monarchies of Spain and Portugal not negotiate a more efficient set of institutions? Alternatively why did the Stuart monarchs in England have to be beheaded or forced from power before better economic institutions could emerge? It seems quite clear that a change to a more efficient set of institutions in Spain and Portugal would not have been possible under the auspices of the absolutist state, and a reduction in the power of the state was certainly inimical to the interest of the crown. In the case of England, Hill (1961a, p. 22) argues directly that the reason that the Tudor and Stuart monarchs were not in favor of efficient economic institutions is because they feared that this would undermine their political power. He notes: “in general the official attitude to industrial advance was hostile, or at best indifferent. It was suspicious of social change and social mobility, the rapid enrichment of capitalists, afraid of the fluctuations of the market and of unemployment, of vagabondage, and social unrest . . . the Elizabethan codes aimed at stabilizing the existing class structure, the location of industry and the flow of labor supply by granting privileges and by putting hindrances in the way of the mobility and the freedom of contract.” The account so far explains why a change in the balance of (de facto) political power in England and the Netherlands led to a set of economic institutions favoring the interests of merchants. But in fact much more happened during the seventeenth century; an entirely new set of political institutions, constitutional regimes, restricting the power of the monarchy, were introduced. The reason why the merchants and the gentry in England (and the merchants in the Netherlands) used their newfound powers for political reform illustrates the dynamics of political power emphasized by our theoretical framework. For example in the case of England, although in 1688 the Parliament might have been strong, it could not be sure that this power would endure. Indeed, the ability to solve the collective action problem and wield de facto power is intrinsically transitory. For instance, the Parliament vanquished James II with the help of a Dutch army, after which they invited William of Orange to take the throne. But how could they anticipate whether or not William would try to assert the absolutist prerogatives that James II had demanded? The way to make transitory power permanent is to embody it into the rules of the game which is exactly what the English Parliament did after 1688. The changes in institutions after 1688 had large and important effects. For instance, in the eighteenth century the English monarchy was able to borrow huge amounts of money because the fiscal control of Parliament guaranteed that it would not default [see Brewer (1988), Stasavage (2003)]. This borrowing has been seen as crucial to the success of the English war machine. Moreover, with Parliament in control of fiscal policy, the crown was no longer able either to raise money through arbitrary taxation, or to grant monopoly rights in exchange for money – issues which had previously been constant sources of friction between the crown and Parliament. Similarly, after 1688, the greater security of property rights in England led to a huge expansion of financial institutions and markets [Neal
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
457
(1990)], which, North and Weingast (1989) argue, laid the institutional foundations for the Industrial Revolution. Of course the English crown was not without some residual power and might have attempted to mount a coup against Parliament to change political institutions back in its favor. This certainly happened in some places, such as in France after 1849 when Louis Napoleon mounted a successful coup to restore absolutist privileges lost in 1848. Nevertheless, changes in political institutions altered the nature of the status quo in significant ways, and therefore, influenced the future distribution of de jure political power. Political institutions are not cast stone, and they can change, but they still create a source of political power more durable than mere de facto power. 9.2. Summary The emergence of constitutional rule in some societies of early modern Europe therefore provides a nice example of how economic institutions, which shape economic outcomes, are determined by political power, which is in turn determined by political institutions and the distribution of resources in society. The Netherlands and England prospered in this period because they had good economic institutions, particularly secure property rights and well developed financial markets. They had these economic institutions because their governments were controlled by groups with a strong vested interest in such economic institutions. These groups wielded political power because of the structure of political institutions, i.e., they received de jure power in the Netherlands after the Dutch Revolt and in England after the Civil War and Glorious Revolution. Moving one step back, we see that political institutions allocated more de jure political power to commercial interests in England and the Netherlands than in France, Spain and Portugal because of major changes in political institutions during the 1600s. These changes took place because commercial interests in England and the Netherlands acquired significant de facto political power as a result of their improving economic fortunes, itself a consequence of the interaction of Atlantic trade and the organization of overseas trade in these countries. Crucially for our framework, these commercial interests used their de facto power to reform (or revolutionize) political institutions so as to acquire de jure political power and solidify their gains. These events, therefore, illustrate the various elements of our theoretical framework. In particular, they show how it is useful to think of political institutions and the distribution of economic resources as the state variables of the dynamic system, which determine the distribution of political power, and via this channel, economic institutions and economic outcomes. Political institutions and the distribution of economic resources are, themselves, endogenous, determined by political power and economic institutions, as exemplified by the fact that the distribution of economic resources changed significantly during the sixteenth century as a result of the new economic opportunities presented by the rise of Atlantic trade, and these changes were crucially influenced by the existing economic institutions (the organization of overseas trade). Furthermore,
458
D. Acemoglu et al.
the change in the balance of political power led to the changes in political institutions through the English Civil War, the Glorious Revolution and the Dutch Revolt. 9.3. Rise of electoral democracy in Britain Our second example, based on Acemoglu and Robinson (2000a, 2001, 2003), is the rise of mass democracy. In the early nineteenth century, European countries were run by small elites. Most had elected legislatures, often descendants of medieval parliaments, but the franchise was highly restricted to males with relatively large amounts of assets, incomes or wealth. However, as the century and the Industrial Revolution progressed, this political monopoly was challenged by the disenfranchised who engaged in collective action to force political change. In response to these developments, the elites responded in three ways. First by using the military to repress the opposition, as in the responses to the revolutions of 1848. Second, by making economic concessions to buy off opposition – this is the standard explanation for the beginnings of the welfare state in Germany under Bismarck. Finally, if neither repression nor concessions were attractive or effective, elites expanded the franchise and gave political power to the previously disenfranchised – they created the precedents of modern democracy. The history of the rise of democracy in Britain is in many ways representative of the experiences of many other European countries. The first important move towards democracy in Britain came with the First Reform Act of 1832. This act removed many of the worst inequities under the old electoral system, in particular the ‘rotten boroughs’ where several members of parliament were elected by very few voters. The 1832 reform also established the right to vote based uniformly on the basis of property and income. The reform was passed in the context of rising popular discontent at the existing political status quo in Britain. By the 1820s the Industrial Revolution was well under way and the decade prior to 1832 saw continual rioting and popular unrest. Notable were the Luddite Riots from 1811–1816, the Spa Fields Riots of 1816, the Peterloo Massacre in 1819, and the Swing Riots of 1830 [see Stevenson (1979) for an overview]. Another catalyst for the reforms was the July revolution of 1830 in Paris. Much of this was led and orchestrated by the new middle-class groups who were being created by the spread of industry and the rapid expansion of the British economy. For example, under the pre-1832 system neither Manchester nor Sheffield had any members of the House of Commons. There is little dissent amongst historians that the motive for the 1832 Reform was to avoid social disturbances [e.g., Lang (1999, p. 36)]. The 1832 Reform Act increased the total electorate from 492,700 to 806,000, which represented about 14.5% of the adult male population. Yet, the majority of British people could not vote, and the elite still had considerable scope for patronage, since 123 constituencies still contained less than 1000 voters. There is also evidence of continued corruption and intimidation of voters until the Ballot Act of 1872 and the Corrupt and Illegal Practices Act of 1883. The Reform Act therefore did not create mass democracy, but rather was designed as
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
459
a strategic concession. In presenting his electoral reform to the British Parliament in 1831, the Prime Minister Earl Grey was well aware that this was a measure necessary to prevent a likely revolution. He argued: “The Principal of my reform is to prevent the necessity for revolution . . . reforming to preserve and not to overthrow.” [Quoted in Evans (1983, p. 212)]. Unsurprisingly therefore, the issue of parliamentary reform was still very much alive after 1832, and it was taken up centrally by the Chartist movement. But as Lee (1994, p. 137) notes “The House of Commons was largely hostile to reform because, at this stage, it saw no need for it”. This had changed by 1867, largely due to a juxtaposition of factors, including the sharp business cycle downturn that caused significant economic hardship and the increased threat of violence. Also significant was the founding of the National Reform Union in 1864 and the Reform League in 1865, and the Hyde Park riots of July 1866 provided the most immediate catalyst. Lang (1999, p. 75) sums up his discussion by saying “The Hyde Park affair, coupled with other violent outbursts, helped to underscore the idea that it would be better to keep the goodwill of the respectable workers than to alienate them”. Reform was initially proposed by the Liberal Prime Minister Russell in 1866 but was defeated by the Conservatives and dissident MP’s. As a result Russell’s government fell, and the Conservatives formed a minority administration with Lord Derby as their leader in the House of Lords, and Disraeli in charge of the House of Commons. It was Disraeli who then constructed a coalition to pass the Second Reform Act in 1867. As a result of these reforms, the total electorate was expanded from 1.36 million to 2.48 million, and working class voters became the majority in all urban constituencies. The electorate was doubled again by the Third Reform Act of 1884, which extended the same voting regulations that already existed in the boroughs (urban constituencies) to the counties (rural constituencies). The Redistribution Act of 1885 removed many remaining inequalities in the distribution of seats and from this point on Britain only had single member electoral constituencies (previously many constituencies had elected two members – the two candidates who gained the most votes). After 1884 about 60% of adult males were enfranchised. Once again social disorder appears to have been an important factor behind the 1884 act. In Britain, the Reform Acts of 1867–1884 were a turning point in the history of the British state. Economic institutions also began to change. In 1871 Gladstone reformed the civil service, opening it to public examination, making it meritocratic. Liberal and Conservative governments introduced a considerable amount of labor market legislation, fundamentally changing the nature of industrial relations in favor of workers. During 1906–1914, the Liberal Party, under the leadership of Asquith and Lloyd George, introduced the modern redistributive state into Britain, including health and unemployment insurance, government financed pensions, minimum wages, and a commitment to redistributive taxation. As a result of the fiscal changes, taxes as a proportion of National Product more than doubled in the 30 years following 1870, and then doubled again. In the meantime, the progressivity of the tax system also increased [Lindert (2004)]. Fi-
460
D. Acemoglu et al.
nally, there is also a consensus amongst economic historians that inequality in Britain fell after the 1870’s [see Lindert (2000, 2004)]. Meanwhile, the education system, which was either primarily for the elite or run by religious denominations during most of the nineteenth century, was opened up to the masses; the Education Act of 1870 committed the government to the systematic provision of universal education for the first time, and this was made free in 1891. The school leaving age was set at 11 in 1893, then in 1899, it increased to 12 and special provisions for the children of needy families were introduced [Mitch (1993)]. As a result of these changes, the proportion of 10-year olds enrolled in school that stood at 40 percent in 1870 increased to 100 percent in 1900 [Ringer (1979, p. 207)]. Finally, a further act in 1902 led to a large expansion in the resources for schools and introduced the grammar schools which subsequently became the foundation of secondary education in Britain. Following the Great War, the Representation of the People Act of 1918 gave the vote to all adult males over the age of 21, and women over the wage of 30 who were ratepayers or married to ratepayers. Ultimately, all women received the vote on the same terms as men in 1928. The measures of 1918 were negotiated during the war and may reflect to some extent a quid pro quo between the government and the working classes who were needed to fight and produce munitions. Nevertheless, Garrard (2002, p. 69) notes “most assumed that, if the system was to survive and ‘contentment and stability prevail’, universal citizenship could not be denied men, perceived to have suffered so much and to have noticed Russia’s Revolution”. Overall, the picture which emerges from British political history is clear. Beginning in 1832, when Britain was governed by the relatively rich, primarily rural aristocracy, a series of strategic concessions were made over an 86 year period. These concessions were aimed at incorporating the previously disenfranchised into politics since the alternative was seen to be social unrest, chaos and possibly revolution. The concessions were gradual because in 1832, social peace could be purchased by buying off the middle classes. Moreover, the effect of the concessions was diluted by the specific details of political institutions, particularly the continuing unrepresentative nature of the House of Lords. Although challenged during the 1832 reforms, the House of Lords provided an important bulwark for the wealthy against the potential of radical reforms emanating from a democratized House of Commons. Later, as the working classes reorganized through the Chartist movement and later through trade unions, further concessions had to be made. The Great War and the fallout from it sealed the final offer of full democracy. Though the pressure of the disenfranchised played less of a role in some reforms than others, and other factors undoubtedly played a role, the threat of social disorder was the main driving force behind the creation of democracy in Britain. The story of the rise of mass democracy that emerges from the British evidence is one where economic and social changes connected with industrialization (for example, rising inequality) and urbanization increased the de facto power of the disenfranchised. In response, they demanded political rights, in particular changes in the political institutions which would allocate future political power to them. These changes in political
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
461
institutions were, in many ways, the direct cause of the changes in economic institutions, in particular, in the labor market, in government policy, in the educational system, with major distributional implications, including the fall in inequality. Why did elites in Britain create a democracy? Our discussion makes it clear that democracy did not emerge from the voluntary acts of an enlightened elite. Democracy was, in many ways, forced on the elite, because of the threat of revolution. Nevertheless, democratization was not the only potential outcome in the face of pressure from the disenfranchised, or even in the face of the threat of revolution. Many other countries faced the same pressures and political elites decided to repress the disenfranchised rather than make concessions to them. This happened with regularity in Europe in the nineteenth century, though by the turn of the twentieth century most had accepted that democracy was inevitable. Repression lasted much longer as the favorite response of elites in Latin America, and it is still the preferred option for current political elites in China or Burma. The problem with repression is that it is costly. Faced with demands for democracy political elites face a trade-off. If they grant democracy, then they lose power over policy and face the prospect of, possibly radical, redistribution. On the other hand, repression risks destroying assets and wealth. In the urbanized environment of nineteenth century Europe (Britain was 70% urbanized at the time of the Second Reform Act), the disenfranchised masses were relatively well organized and therefore difficult to repress. Moreover, industrialization had led to an economy based on physical, and increasing human, capital. Such assets are easily destroyed by repression and conflict, making repression an increasingly costly option for elites. In contrast, in predominantly agrarian societies like many parts of Latin America earlier in the century or current-day Burma, physical and human capital are relatively unimportant and repression is easier and cheaper. Moreover, not only is repression cheaper in such environments, democracy is potentially much worse for the elites because of the prospect of radical land reform. Since physical capital is much harder to redistribute, elites in Western Europe found the prospect of democracy much less threatening. Faced with the threat of revolt and social chaos, political elites may also attempt to avoid giving away their political power by making concessions, such as income redistribution or other policies that favor non-elites and the disenfranchised. The problem with concessions however is their credibility, particularly when de facto power is transitory. For example, if a crisis, such as a harvest failure or business cycle recession creates a window of opportunity to solve the collective action problem and challenge the existing regime, the elites would like to respond with the promise of policy concessions. Yet windows of opportunity disappear and it is difficult to sustain collective action which entails people protesting in the streets and being away from their families and jobs. Thus collective action quickly dissipates and once it does so, the government has an incentive to renege on its promise of concessions. The promise of concessions, which people know to be non-credible is unlikely to defuse collective action. Hence, Acemoglu and Robinson (2000a, 2001, 2003) argue that democratization occurred as a way of making credible commitments to the disenfranchised. Democratization was a credible commitment to future redistribution, because it reallocated de jure political power away from
462
D. Acemoglu et al.
the elites to the masses. In democracy, the poorer segments of the society would be more powerful and could vote, in other words, could use their de jure political power, to implement economic institutions and policies consistent with their interests. Therefore, democratization was a way of transforming the transitory de facto power of the disenfranchised poor into more durable de jure political power. 9.4. Summary The emergence of mass democracy is another example illustrating our theory of institutions. Into the nineteenth century, economic institutions, particularly in the labor market, disadvantaged the poor. For example, trade unions were illegal and as late as the 1850 in Britain workers trying to organize a union could be shipped to the penal colony in Tasmania, Australia. The poor could not alter economic institutions in their favor because, being disenfranchised, they had little de jure political power, and also limited de facto power because they were often unable to solve their collective action problems. However, changes in the structure of society and the economy during the early nineteenth century altered the balance of political power, in particular making the exercise of de facto power by the politically disenfranchised much easier [Tilly (1995) and Tarrow (1998) document the changing qualitative nature of collective action over this period]. The rise in the de facto political power of the poor necessitated a change in political institutions in their favor to defuse the threat of revolution. This was to tilt the future allocation of de jure political power, and consequently to ensure future economic institutions and policies consistent with their interests. Whether or not increases in de facto power translated into democracy depended on a number of factors, in particular how difficult and costly it was for elites to use repression to counter the increase in the power of the masses, and how costly the prospect of democracy was. The changes in political institutions that occurred with democracy had profound implications for economic institutions. In the case of Britain, the period after the Second Reform Act of 1867 led the British state to commit itself to providing universal education and it also led to radical changes in labor market institutions allowing trade unions to form legally for the first time and increasing the bargaining power of labor. Hence economic institutions changed radically in favor of those newly endowed with de jure political power, mostly the relatively poor. This is in fact a relatively general result of democratization. Democracy enfranchises the poor, and the poor are able to use democracy to tilt economic institutions and the distribution of income in society in their favor [Li, Squire and Zou (1998), Rodrik (1999)]. The emergence of democracy in the nineteenth-century Europe therefore also illustrates the workings of our theoretical framework. In particular, it shows how political institutions determine economic institutions and policies, and thus the distribution of resources, and it shows how political institutions change, especially in response to an imbalance of de facto political power, as a credible way of influencing the future allocation of de jure political power.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
463
10. Future avenues In this chapter we have proposed a framework for thinking about why some countries grow faster and are richer than others. We emphasized, following North and Thomas (1973), that most economic growth theory focuses only on proximate determinants of prosperity. Although this body of work has been useful in helping us understand the mechanics of growth, it fails to provide a satisfactory account of why some countries grow while others do not. Even more recent analyses which have emphasized market institutions and imperfections, and even political economy, have not provided convincing explanations for why countries differ in their equilibrium set of institutions. A major research goal must now be to get beyond the neoclassical growth model and its modern extensions, and search for the deeper causes, i.e., the fundamental determinants of growth. We argued that the available evidence is consistent with the view that whether or not a society grows depends on how its economy is organized – on its economic institutions. We then proposed the outlines of a theory of institutions and illustrated it through a series of historical examples. We emphasized that a theory of why different countries have different economic institutions must be based on politics, on the structure of political power, and the nature of political institutions. Much remains to be done. First, the framework we outlined was largely verbal rather than mathematical, and thus, by its very nature, not fully specified. Constructing formal models incorporating and extending these ideas is the most important task ahead. Although some of our past work [e.g., Acemoglu and Robinson (2000a, 2001), Acemoglu (2003b)] formalizes parts of this framework, the full model has not been developed yet. There are also many important issues left out of our framework, which appear to offer fruitful areas for future research. First, though we know that institutions, both economic and political, persist for long periods of time, often centuries (and sometimes millennia), we do not as yet have a satisfactory understanding of the mechanisms through which institutions persist. Second, and closely related, although institutions do generally persist, sometimes they change. We have important examples of societies which have radically changed their political and economic institutions. Some do so for internal reasons, such as France after the Revolution of 1789, and some do because of external pressures such as Japan after the Meiji restoration or Russia after the Crimean War. The important point here is that both institutional persistence and institutional change are equilibrium outcomes. Approaches positing institutional persistence as a matter of fact, and then thinking of institutional changes as unusual events will not be satisfactory. Both phenomena have to be analyzed as part of the same dynamic equilibrium framework. One type of institutional change, consistent with the examples we discussed in this chapter, takes place when those who benefit from the existing set of institutions are forced to accept change, either because they are the losers in a process of fighting or because of the threat of internal revolution (another possibility is that they might accept
464
D. Acemoglu et al.
change because of the threat of external invasion). However, institutional change can also take place because the set of economic institutions that is optimal for a particular group with political power may vary over time as the state variables in the system and economic opportunities evolve. One example may be the end of slavery in the British Empire and another may be the economic and political changes introduced by Mikhail Gorbachev in the Soviet Union in the 1980s. We need more research on the dynamic mechanisms at work [see Tornell (1997) for a model of such a process]. Finally, it is important to understand the role of policy and interventions in changing the institutional equilibrium. Though social science research is of intrinsic interest, one would hope that a convincing fundamental theory of comparative growth based on institutions would lead to policy conclusions that would help us improve the institutions and thus the lives and welfare of people in poor countries. It should be obvious that, at the moment, we are a long way from being in a position to draw such conclusions. In a world where political choices are made rationally and are endogenous to the structure of institutions, which are themselves ultimately endogenous, giving policy advice is a conceptually complex issue [see Acemoglu et al. (2003) for reflections on this]. Recognizing our current ignorance on this topic in no way diminishes its importance, and its role as the Holy Grail of political economy research, however. And we believe that better and empirically more realistic theoretical frameworks in the future will take us closer to this Holy Grail.
Acknowledgements We thank the editors for their outstanding patience and Philippe Aghion, Robert Barro, Avinash Dixit, Leopoldo Fergusson, Herschel Grossman, Pablo Querubín and Barry Weingast for their helpful suggestions.
References Acemoglu, D. (1995). “Reward structures and the allocation of talent”. European Economic Review 39, 17– 33. Acemoglu, D. (1997). “Training and innovation in an imperfect labor market”. Review of Economic Studies 64, 445–464. Acemoglu, D. (2003a). “Why not a Political Coase Theorem?”. Journal of Comparative Economics 31, 620– 652. Acemoglu, D. (2003b). “The form of property rights: Oligarchic versus democratic societies”. NBER Working Paper #10037. Acemoglu, D., Aghion, P., Zilibotti, F. (2002). “Distance to frontier, selection, and economic growth”. NBER Working Paper #9066. Acemoglu, D., Johnson, S., Robinson, J.A. (2001). “The colonial origins of comparative development: An empirical investigation”. American Economic Review 91 (5), 1369–1401. Acemoglu, D., Johnson, S., Robinson, J.A. (2002). “Reversal of fortune: Geography and institutions in the making of the modern world income distribution”. Quarterly Journal of Economics 118, 1231–1294.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
465
Acemoglu, D., Johnson, S., Robinson, J.A. (2005). “The rise of Europe: Atlantic trade, institutional change and economic growth.” American Economic Review, in press. Acemoglu, D., Robinson, J.A. (2000a). “Why did the West extend the franchise? Democracy, inequality and growth in historical perspective”. Quarterly Journal of Economics 115, 1167–1199. Acemoglu, D., Robinson, J.A. (2000b). “Political losers as a barrier to economic development”. American Economic Review 90, 126–130. Acemoglu, D., Robinson, J.A. (2001). “A theory of political transitions”. American Economic Review 91, 938–963. Acemoglu, D., Robinson, J.A. (2002). “Economic backwardness in political perspective”. NBER Working Paper #8831. Acemoglu, D., Robinson, J.A. (2003). Economic Origins of Dictatorship and Democracy, forthcoming book to be published by Cambridge University Press. Acemoglu, D., Zilibotti, F. (1997). “Was Prometheus unbound by chance? Risk, diversification and growth”. Journal of Political Economy 105, 709–751. Acemoglu, D., Johnson, S., Robinson, J.A., Thaicharoen, Y. (2003). “Institutional causes, macroeconomic symptoms: Volatility, crises and growth”. Journal of Monetary Economics 50, 49–123. Ades, A., Verdier, T. (1996). “The rise and fall of elites: A theory of economic development and social polarization in rent-seeking societies”. CEPR Discussion Paper #1495. Aghion, P., Howitt, P.W. (1992). “A model of growth through creative destruction”. Econometrica 60, 323– 351. Aghion, P., Howitt, P.W. (1994). “Growth and unemployment”. Review of Economic Studies 61, 477–494. Aghion, P., Howitt, P.W. (2005). “Growth with quality-improving innovations: An integrated framework”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1A. Elsevier, Amsterdam. Aghion, P., Harris, C., Howitt, P.W., Vickers, J. (2001). “Competition, imitation, and growth with step-by-step innovation”. Review of Economic Studies 68, 467–492. Alesina, A., Perotti, R. (1996). “Income distribution, political instability and investment”. European Economic Review 40, 1203–1225. Alesina, A., Rodrik, D. (1994). “Distributive politics and economic growth”. Quarterly Journal of Economics 109, 465–490. Allen, R.C. (1982). “The efficiency and distributional consequences of eighteenth century enclosures”. Economic Journal 92, 937–953. Allen, R.C. (1992). Enclosure and the Yeoman. Oxford University Press, New York. Aston, T.H., Philpin, C.H.E. (Eds.) (1985). The Brenner Debate: Agrarian Class Structure and Economic Development in Pre-industrial Europe. Cambridge University Press, New York. Azariadis, C., Stachurski, J. (2005). “Poverty traps”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1A. Elsevier, Amsterdam. Bairoch, P. (1988). Cities and Economic Development: From the Dawn of History to the Present. University of Chicago Press, Chicago. Bairoch, P., Batou, J., Chèvre, P. (1988). La Population des villes Europeenees de 800 a 1850: Banque de Données et Analyse Sommaire des Résultats. Centre D’histoire Économique Internationale de l’Uni. de Genève, Libraire Droz, Geneva. Banerjee, A., Duflo, E. (2005). “Growth theory through the lens of development economics”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1A. Elsevier, Amsterdam. Banerjee, A., Newman, A.F. (1993). “Occupational choice and the process of development”. Journal of Political Economy 101, 274–298. Banfield, E.C. (1958). The Moral Basis of a Backward Society. University of Chicago Press, Chicago. Bardhan, P., Udry, C. (1999). Development Microeconomics. Oxford University Press, New York. Barro, R.J. (1997). The Determinants of Economic Growth: A Cross-Country Empirical Study. MIT Press, Cambridge. Barro, R.J., McCleary, R. (2003). “Religion and economic growth”. NBER Working Paper #9682. Barzel, Y. (2001). A Theory of the State: Economic Rights, Legal Rights, and the Scope of the State. Cambridge University Press, New York.
466
D. Acemoglu et al.
Bates, R.H. (1981). Markets and States in Tropical Africa. University of California Press, Berkeley, CA. Bates, R.H. (1989). Beyond the Miracle of the Market. Cambridge University Press, New York. Bates, R.H. (1997). Open Economy Politics. Princeton University Press, Princeton. Becker, G.S. (1958). “Competition and democracy”. Journal of Law and Economics 1, 105–109. Benhabib, J., Rustichini, A. (1996). “Social conflict and growth”. Journal of Economic Growth 1, 125–142. Berman, B.J., Lonsdale, J. (1992). Unhappy Valley. James Currey, London. Besley, T.F., Coate, S.T. (1998). “Sources of inefficiency in a representative democracy: A dynamic analysis”. American Economic Review 88, 139–156. Bloom, D.E., Sachs, J.D. (1998). “Geography, demography, and economic growth in Africa”. Brookings Papers on Economic Activity 2, 207–295. Blum, J. (1943). “Transportation and industry in Austria, 1815–1848”. Journal of Modern History 15, 24–38. Bourguignon, F., Verdier, T. (2000). “Oligarchy, democracy, inequality and growth”. Journal of Development Economics 62, 285–313. Brenner, R. (1973). “The Civil War politics of Londons merchant community”. Past and Present 58, 53–107. Brenner, R. (1976). “Agrarian class structure and economic development in preindustrial Europe”. Past and Present 70, 30–75. Brenner, R. (1982). “Agrarian roots of European capitalism”. Past and Present 97, 16–113. Brenner, R. (1993). Merchants and Revolution: Commercial Change, Political Conflict, and London’s Overseas Traders, 1550–1653. Princeton University Press, Princeton. Brewer, J. (1988). The Sinews of Power: War, Money, and the English State, 1688–1783. Harvard University Press, Cambridge. Brunton, D., Pennington, D.H. (1954). Members of the Long Parliament. Allen and Unwin, London. Cameron, R. (1993). A Concise Economic History of the World. Oxford University Press, New York. Cantor, N.F. (2001). In the Wake of the Plague: The Black Death and the World it Made. The Free Press, New York. Cardosso, F.H., Faletto, E. (1979). Dependency and Development in Latin America. University of California Press, Berkeley, CA. Cass, D. (1965). “Optimum growth in an aggregate model of capital accumulation”. Review of Economic Studies 32, 233–240. Chandler, T. (1987). Four Thousand Years of Urban Growth: An Historical Census. St. David’s University Press, Lewiston, NY. Charlesworth, A. (1983). An Atlas of Rural Protest in Britain 1548–1900. Croon Helm, London. Coase, R.H. (1937). “The nature of the firm”. Economica 3, 386–405. Coase, R.H. (1960). “The problem of social cost”. Journal of Law and Economics 3, 1–44. Coatsworth, J.H. (1993). “Notes on the comparative economic history of Latin America and the United States”. In: Bernecker, W.L., Werner Tobler, H. (Eds.), Development and Underdevelopment in America: Contrasts in Economic Growth in North and Latin America in Historical Perspective. Walter de Gruyter, New York. Cumings, B. (2004). North Korea: Another Country. The New Press, New York. Curtin, P.D. (1989). Death by Migration: Europe’s Encounter with the Tropical World in the Nineteenth Century. Cambridge University Press, New York. Curtin, P.D. (1998). Disease and Empire: The Health of European Troops in the Conquest of Africa. Cambridge University Press, New York. Davis, R. (1973a). The Rise of the Atlantic Economies. Cornell University Press, Ithaca. Davis, R. (1973b). English Overseas Trade 1500–1700. Macmillan, London. Demsetz, H. (1967). “Toward a theory of property right”. American Economic Review 57, 61–70. de Vries, J., van der Woude, A. (1997). The First Modern Economy: Success, Failure, and Perseverance of the Dutch Economy, 1500–1815. Cambridge University Press, New York. Diamond, J.M. (1997). Guns Germs and Steel: The Fate of Human Societies. W.W. Norton & Co., New York. Dixit, A.K. (1996). The Making of Economic Policy: A Transaction Cost Politics Perspective. MIT Press, Cambridge.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
467
Dixit, A.K. (2004). Lawlessness and Economics: Alternative Modes of Governance. Princeton University Press, Princeton. Dixit, A.K., Londregan, J.B. (1995). “Redistributive politics and economic efficiency”. American Political Science Review 89, 856–866. Djankov, S., La Porta, R., Lopez-de-Silanes, F., Shleifer, A. (2002). “The regulation of entry”. Quarterly Journal of Economics 117, 1–37. Djankov, S., La Porta, R., Lopez-de-Silanes, F., Shleifer, A. (2003). “Courts”. Quarterly Journal of Economics 118, 453–517. Dobb, M.H. (1948). Studies in the Development of Capitalism. Cambridge University Press, Cambridge, UK. Domar, E. (1970). “The causes of slavery or serfdom: A hypothesis”. Journal of Economic History 30, 18–32. Durlauf, S.N. (1993). “Nonergodic economic growth”. Review of Economic Studies 60, 349–366. Durlauf, S.N., Fafchamps, M. (2004). “Empirical studies of social capital: A critical survey”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1A. Elsevier, Amsterdam. Edgerton, R.B. (1992). Sick Societies: Challenging the Myth of Primitive Harmony. Free Press, New York. Eggimann, G. (1999). La Population des Villes des Tiers-Mondes, 1500–1950. Centre d’histoire économique Internationale de l’Uni. de Genève, Libraire Droz, Geneva. Engerman, S.L., Sokoloff, K.L. (1997). “Factor endowments, institutions, and differential growth paths among new world economies”. In: Haber, S. (Ed.), How Latin America Fell Behind. Stanford University Press, Stanford, CA. Ertman, T. (1997). Birth of the Leviathan: Building States and Regimes in Medieval and Early Modern Europe. Cambridge University Press, New York. Evans, E.J. (1983). The Forging of the Modern State: Early Industrial Britain, 1783–1870. Longman, New York. Farrell, J. (1987). “Information and the Coase Theorem”. Journal of Economic Perspectives 1, 113–129. Frank, A.G. (1978). Dependent Accumulation and Underdevelopment. Macmillan, London. Freudenberger, H. (1967). “State intervention as an obstacle to economic growth in the Hapsburg Monarchy”. Journal of Economic History 27, 493–509. Galor, O., Zeira, J. (1993). “Income distribution and macroeconomics”. Review of Economic Studies 40, 35–52. Garrard, J. (2002). Democratization in Britain: Elites, Civil Society and Reform since 1800. Basingstoke, Palgrave. Gerschenkron, A. (1970). Europe in the Russian Mirror: Four Lectures in Economic History. Cambridge University Press, Cambridge, UK. Glaeser, E.L., Shleifer, A. (2002). “Legal origins”. Quarterly Journal of Economics 117, 1193–1230. Graves, M.A.R. (2001). The Parliaments of Early Modern Europe. Longman, New York. Gregory, P.R. (1991). “The role of the state in promoting economic development: The Russian case and its general implications”. In: Sylla, R., Toniolo, G. (Eds.), Patterns of European Industrialization: The Nineteenth Century. Routledge, New York. Greif, A. (1994). “Cultural beliefs and the organization of society: A historical and theoretical reflection on collectivist and individualist societies”. Journal of Political Economy 102, 912–950. Gross, N. (1973). “The industrial revolution in the Hapsburg monarchy, 1750–1914”. In: Cipolla, C.M. (Ed.), The Fontana Economic History of Europe, vol. 4. Fontana Books, London. Grossman, G.M., Helpman, E. (1991). Innovation and Growth in the Global Economy. MIT Press, Cambridge. Grossman, H., Kim, M. (1995). “Swords or ploughshares? A theory of the security of claims to property rights”. Journal of Political Economy 103, 1275–1288. Grossman, H., Kim, M. (1996). “Predation and accumulation”. Journal of Economic Growth 1, 333–350. Grossman, H., Noh, S.J. (1994). “Proprietary public finance and economic welfare”. Journal of Public Economics 53, 187–204. Grossman, S.J., Hart, O.D. (1986). “The costs and benefits of ownership: A theory of vertical and lateral integration”. Journal of Political Economy 94, 691–719. Gutierrez, H. (1986). “La mortalite des eveques Latino-Americains aux XVIIe et XVIII siecles”. Annales de Demographie Historique, 29–39.
468
D. Acemoglu et al.
Haber, S.H. (2001). “Political institutions and banking systems: Lessons from the economic histories of Mexico and the United States, 1790–1914”. Unpublished. Department of Political Science, Stanford University. Haber, S.H., Maurer, N. (2004). “Related lending and economic performance: Evidence from Mexico”. Unpublished. Department of Political Science, Stanford University. Hall, R.E., Jones, C.I. (1999). “Why do some countries produce so much more output per worker than others?”. Quarterly Journal of Economics 114, 83–116. Hardin, R. (1995). All For One. Princeton University Press, Princeton. Hart, O.D. (1995). Firms, Contracts, and Financial Structure. Oxford University Press, New York. Herbst, J.I. (1990). State Politics in Zimbabwe. University of California Press, Berkeley. Hicks, J.R. (1969). A Theory of Economic History. Oxford University Press, New York. Hill, C. (1961a). The Century of Revolution, 1603–1714. W.W. Norton & Co., New York. Hill, C. (1961b). “Protestantism and the rise of capitalism”. In: Fisher, F.J. (Ed.), Essays in the Economic and Social History of Tudor and Stuart England. Cambridge University Press, Cambridge. Hill, C. (1969). From Reformation to Industrial Revolution 1530–1780. Penguin Books, Baltimore. Hilton, R. (1981). Bond Men Made Free. Routledge, Oxford. Hirshleifer, J. (2001). The Dark Side of the Force: Economic Foundations of Conflict Theory. Cambridge University Press, New York. Horowitz, D.L. (1991). A Democratic South Africa? Constitutional Engineering in Divided Societies. University of California Press, Berkeley. Israel, J.I. (1989). Dutch Primacy in World Trade, 1585–1740. The Clarendon Press, Oxford. Israel, J.I. (1995). The Dutch Republic: Its Rise, Greatness and Fall 1477–1806. Oxford University Press, New York. Jones, E.L. (1981). The European Miracle: Environments, Economies, and Geopolitics in the History of Europe and Asia. Cambridge University Press, New York. Keeler, M. (1954). The Long Parliament, 1640–1641; A Biographical Study of its Members. American Philosophical Society, Philadelphia. Keyssar, A. (2000). The Right to Vote: The Contested History of Democracy in the United States. Basic Books, New York. Killick, T. (1978). Development Economics in Action: A Study of Economic Policies in Ghana. Heinemann, London. Knack, S., Keefer, P. (1995). “Institutions and economic performance: Cross-country tests using alternative measures”. Economics and Politics 7, 207–227. Knack, S., Keefer, P. (1997). “Does social capital have an economic impact? A cross-country investigation”. Quarterly Journal of Economics 112, 1252–1288. Koopmans, T.C. (1965). “On the concept of optimal economic growth”. In: The Economic Approach to Development Planning. North-Holland, Amsterdam. Krugman, P.R., Venables, A.J. (1995). “Globalization and the inequality of nations”. Quarterly Journal of Economics 110, 857–880. Krusell, P., Ríos-Rull, J.-V. (1996). “Vested interests in a theory of stagnation and growth”. Review of Economic Studies 63, 301–330. Krusell, P., Ríos-Rull, J.-V. (1999). “On the size of government: Political economy in the neoclassical growth model”. American Economic Review 89, 1156–1181. Kupperman, K.O. (1993). Providence Island, 1630–1641: The other Puritan Colony. Cambridge University Press, New York. Kuznets, S. (1968). Towards a Theory of Economic Growth. Yale University Press, New Haven, CT. Landes, D.S. (1998). The Wealth and Poverty of Nations: Why Some Are so Rich and Some so Poor. Norton & Co, New York. Lang, S. (1999). Parliamentary Reform, 1785–1928. Routledge, New York. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R. (1998). “Law and finance”. Journal of Political Economy 106, 1113–1155.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
469
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R. (1999). “The quality of government”. Journal of Law, Economics and Organization 15, 222–279. Lee, S.J. (1994). Aspects of British Political History, 1815–1914. Routledge, New York. Levine, R. (2005). “Finance and growth: Theory and evidence”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1A. Elsevier, Amsterdam. Lewis, W.A. (1978). The Emergence of the International Economic Order. Princeton University Press, Princeton. Li, H., Squire, L., Zou, H.-f. (1998). “Explaining international and intertemporal variations in income inequality”. Economic Journal 108, 26–43. Lichbach, M.I. (1995). The Rebel’s Dilemma. University of Michigan Press, Ann Arbor. Lindert, P.H. (2000). “Three centuries of inequality in Britain and America”. In: Atkinson, A.B., Bourguignon, F. (Eds.), Handbook of Income Distribution. North-Holland, Amsterdam. Lindert, P.H. (2004). Growing Public: Social Spending and Economics Growth since the Eighteenth Century, vols.1, 2. Cambridge University Press, Cambridge, MA. Lucas, R.E. (1988). “On the mechanics of economic development”. Journal of Monetary Economics 22, 3–42. Maddison, A. (2001). The World Economy: A Millennial Perspective. Development Centre of the Organization for Economic Cooperation and Development, OECD, Paris. Marshall, A. (1890). Principles of Economics. Macmillan, London. 1949. Maurer, N. (2002). The Power and the Profits: The Mexican Financial System, 1876–1932. Stanford University Press, Stanford. McDaniel, T. (1991). Autocracy, Modernization and Revolution in Russia and Iran. Princeton University Press, Princeton. McEvedy, C., Jones, R. (1978). Atlas of World Population History. Facts on File, New York. Mitch, D. (1993). “The role of human capital in the first industrial revolution”. In: Mokyr, J. (Ed.), The British Industrial Revolution: An Economic Perspective. Westview Press, San Francisco. Mokyr, J. (1990). The Lever of Riches: Technological Creativity and Economic Progress. Oxford University Press, New York. Montesquieu, C.S. (1748). The Spirit of the Laws. Cambridge University Press, New York. 1989. Moore, B. Jr. (1966). Social Origins of Dictatorship and Democracy: Lord and Peasant in the Making of the Modern World. Beacon Press, Boston. Morgan, E.S. (1975). American Slavery, American Freedom: The Ordeal of Colonial Virginia. W.W. Norton & Co., New York. Mosse, W.E. (1958). Alexsandr II and the Modernization of Russia. University of London Press, London, UK. Mosse, W.E. (1992). An Economic History of Russia, 1856–1914. I.B. Taurus Press, London, UK. Murphy, K.M., Shleifer, A., Vishny, R.W. (1989a). “Industrialization and the big push”. Journal of Political Economy 97, 1003–1026. Murphy, K.M., Shleifer, A., Vishny, R.W. (1989b). “Income distribution, market size and industrialization”. Quarterly Journal of Economics 104, 537–564. Murphy, K.M., Shleifer, A., Vishny, R.W. (1991). “The allocation of talent: Implications for growth”. Quarterly Journal of Economics 106, 503–530. Myrdal, G. (1968). Asian Drama; An Inquiry into the Poverty of Nations, vols. 1–3. Twentieth Century Fund, New York. Neal, L. (1990). The Rise of Financial Capitalism: International Capital Markets in the Age of Reason. Cambridge University Press, New York. Newton, A.P. (1914). The Colonizing Activities of the English Puritans. Yale University Press, New Haven. North, D.C. (1981). Structure and Change in Economic History. W.W. Norton & Co., New York. North, D.C. (1990). Institutions, Institutional Change, and Economic Performance. Cambridge University Press, New York. North, D.C., Thomas, R.P. (1973). The Rise of the Western World: A New Economic History. Cambridge University Press, Cambridge, UK.
470
D. Acemoglu et al.
North, D.C., Summerhill, W., Weingast, B.R. (2000). “Order, disorder, and economic change: Latin America versus North America”. In: de Mesquita, B.B., Root, H.L. (Eds.), Governing for Prosperity. Yale University Press, New Haven. North, D.C., Weingast, B.R. (1989). “Constitutions and commitment: Evolution of institutions governing public choice in seventeenth century England”. Journal of Economic History 49, 803–832. Olson, M.C. (1982). The Rise and Decline of Nations: Economic Growth, Stagflation, and Economic Rigidities. Yale University Press, New Haven and London. Olson, M.C. (1993). “Dictatorship, democracy and development”. American Political Science Review 87, 567–575. Olson, M.C. (2000). Power and Prosperity: Outgrowing Communist and Capitalist Dictatorships. Basic Books, New York. Overton, M. (1996). Agricultural Revolution in England: The Transformation of the Agrarian Economy 1500– 1850. Cambridge University Press, New York. Parente, S., Prescott, E.C. (1999). “Monopoly rights as barriers to riches”. American Economic Review 89, 1216–1233. Parente, S., Prescott, E.C. (2005). “A unified theory of the evolution of international income levels”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1B. Elsevier, Amsterdam. Parry, J.H. (1966). The Spanish Seaborne Empire. University of California Press, Berkeley. Perotti, R. (1993). “Political equilibrium, income distribution and growth”. Review of Economic Studies 61, 755–776. Persson, T., Tabellini, G. (1994). “Is inequality harmful for growth?”. American Economic Review 84, 600– 621. Piketty, T. (1995). “Social mobility and redistributive politics”. Quarterly Journal of Economics 100, 551– 584. Pincus, S. (1998). “Neither Machiavellian moment nor possessive individualism: Commercial society an the defenders of the English commonwealth”. American Historical Review 103, 705–736. Pincus, S. (2001). “From holy cause to economic interest: The study of population and the invention of the state”. In: Houston, A., Pincus, S. (Eds.), A Nation Transformed: England after the Restoration. Cambridge University Press, New York. Pincus, S. (2002). “Civic republicanism and political economy in an age of revolution: Law, politics, and economics in the revolution of 1688–89”. Unpublished. Department of History, University of Chicago. Pirenne, H. (1937). Economic and Social History of Medieval Europe. Harcourt, Brace and Company, New York. Pissarides, C. (2000). Equilibrium Unemployment Theory, second edn. MIT Press, Cambridge. Postan, M.M. (1937). “The chronology of labour services”. Transactions of the Royal Historical Society 20. Postan, M.M. (1966). “Medieval agrarian society in its prime: England”. In: Postan, M.M. (Ed.), The Cambridge Economic History of Europe. Cambridge University Press, London. Powell, R. (2004). “The inefficient use of power: Costly conflict with complete information”. American Political Science Review 98, 231–241. Putnam, R.D., Leonardi, R., Nanetti, R.Y. (1993). Making Democracy Work: Civic Traditions in Modern Italy. Princeton University Press, Princeton. Randall, A. (1991). Before the Luddites: Custom, Community, and Machinery in the English Woollen Industry, 1776–1809. Cambridge University Press, New York. Ray, D. (1998). Development Economics. Princeton University Press, Princeton. Reynolds, A. (1999). Electoral Systems and Democratization in Southern Africa. Oxford University Press, New York. Ringer, F. (1979). Education and Society in Modern Europe. University of Indiana Press, Bloomington. Robinson, J.A. (1998). “Theories of bad policy”. Journal of Policy Reform 3, 1–46. Rodney, W. (1972). How Europe Underdeveloped Africa. Howard University Press, Washington, DC. Rodrik, D. (1999). “Democracies pay higher wages”. Quarterly Journal of Economics CXIV, 707–738. Romer, D. (2003). “Misconceptions and political outcomes”. Economic Journal 113, 1–20.
Ch. 6: Institutions as a Fundamental Cause of Long-Run Growth
471
Romer, P.M. (1986). “Increasing returns and long-run growth”. Journal of Political Economy 94, 1002–1037. Romer, P.M. (1990). “Endogenous technical change”. Journal of Political Economy 98, 71–102. Rosenstein-Rodan, P. (1943). “Problems of industrialization in Eastern and South-Eastern Europe”. Economic Journal 53, 202–211. Ross, J.I., Gurr, T.R. (1989). “Why terrorism subsides: A comparative study of Canada and the United States”. Comparative Politics 21, 405–426. Sachs, J.D. (2000). “Notes on a new sociology of economic development”. In: Harrison, L.E., Huntington, S.P. (Eds.), Culture Matters: How Values Shape Human Progress. Basic Books, New York. Sachs, J.D. (2001). “Tropical underdevelopment”. NBER Working Paper #8119. Saint-Paul, G., Verdier, T. (1993). “Education, democracy, and growth”. Journal of Development Economics 42, 399–407. Schattschneider, E.E. (1935). Politics, Pressures and the Tariff; a Study of Free Private Enterprise in Pressure Politics, as Shown in the 1929–1930 Revision of the Tariff. Prentice-Hall, New York. Scott, J.C. (2000). “The moral economy as an argument and as a fight”. In: Randall, A., Charlesworth, A. (Eds.), Moral Economy and Popular Protest: Crowds, Conflict and Authority. MacMillan, London. Skaperdas, S. (1992). “Cooperation, conflict, and power in the absence of property rights”. American Economic Review 82, 720–739. Smith, A. (1776). The Wealth of Nations, vols.1, 2. Penguin Classics, London. 1999. Solow, R.M. (1956). “A contribution to the theory of economic growth”. Quarterly Journal of Economics 70, 65–94. Stasavage, D. (2003). Public Debt and the Birth of the Democratic State: France and Great Britain, 1688– 1789. Cambridge University Press, New York. Stevenson, J. (1979). Popular Disturbances in England, 1700–1870. Longman, New York. Tarrow, S. (1991). “Aiming at a moving target: Social science and the recent rebellions in Eastern Europe”. PS: Political Science and Politics 24, 12–20. Tarrow, S. (1998). Power in Movement: Social Movements and Contentious Politics, second edn. Cambridge University Press, New York. Tawney, R.H. (1926). Religion and the Rise of Capitalism: A Historical Study. J. Murray, London. Tawney, R.H. (1941). “The rise of the gentry, 1558–1640”. Economic History Review 11, 1–38. Thomis, M.I. (1970). The Luddites; Machine-Breaking in Regency England. David & Charles, Newton Abbot. Tilly, C. (1995). Popular Contention in Britain, 1758–1834. Harvard University Press, Cambridge. Tornell, A. (1997). “Economic growth and decline with endogenous property rights”. Journal of Economic Growth 2, 219–250. Tornell, A., Velasco, A. (1992). “Why does capital flow from poor to rich countries? The tragedy of the commons and economic growth”. Journal of Political Economy 100, 1208–1231. Townsend, R.M. (1993). The Medieval Village Economy: A Study of the Pareto Mapping in General Equilibrium Models. Princeton University Press, Princeton. Vargas Llosa, M. (1989). The Storyteller. Farrar, Straus, Giroux, New York. Veitch, J.M. (1986). “Repudiations and confiscations by the medieval state”. Journal of Economic History 46, 31–36. Véliz, C. (1994). The New World of the Gothic Fox: Culture and Economy in English and Spanish America. University of California Press, Berkeley. Wallerstein, I.M. (1974–1980). The Modern World-System, vols. 1–3. Academic Press, New York. Weber, M. (1930). The Protestant Ethic and the Spirit of Capitalism. Allen and Unwin, London. Weber, M. (1958). The Religion of India. Free Press, Glencoe. Weingast, B.R. (1997). “The political foundations of democracy and the rule of law”. American Political Science Review 91, 245–263. Weingast, B.R. (1998). “Political stability and Civil War: Institutions, commitment, and American democracy”. In: Bates, R., Greif, A., Levi, M., Rosenthal, J.-L., Weingast, B.R. (Eds.), Analytic Narratives. Princeton University Press, Princeton. Wiarda, H.J. (2001). The Soul of Latin America: The Cultural and Political Tradition. Yale University Press, New Haven.
472
D. Acemoglu et al.
Williams, E.E. (1944). Capitalism and Slavery. University of North Carolina Press, Chapel Hill. Williamson, O. (1985). The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. Free Press, New York. Wittman, D. (1989). “Why democracies produce efficient results”. Journal of Political Economy 97, 1395– 1424.
Chapter 7
GROWTH THEORY THROUGH THE LENS OF DEVELOPMENT ECONOMICS ABHIJIT V. BANERJEE AND ESTHER DUFLO MIT, Department of Economics, 50 Memorial Drive, Cambridge, MA 02142, USA e-mails:
[email protected];
[email protected]
Contents Abstract Keywords 1. Introduction: neo-classical growth theory 1.1. The aggregate production function 1.2. The logic of convergence
2. Rates of return and investment rates in poor countries 2.1. Are returns higher in poor countries? 2.1.1. Physical capital 2.1.2. Human capital 2.1.3. Taking stock: returns on capital 2.2. Investment rates in poor countries 2.2.1. Is investment higher in poor countries? 2.2.2. Does investment respond to rates of return? 2.2.3. Taking stock: investment rates
3. Understanding rates of return and investment rates in poor countries: aggregative approaches 3.1. 3.2. 3.3. 3.4.
Access to technology and the productivity gap Human capital externalities Coordination failure Taking stock
4. Understanding rates of return and investment rates in poor countries: nonaggregative approaches 4.1. Government failure 4.1.1. Excessive intervention 4.1.2. Lack of appropriate regulations: property rights and legal enforcement 4.2. The role of credit constraints 4.3. Problems in the insurance markets 4.4. Local externalities
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01007-5
474 474 475 475 477 479 479 479 484 491 493 493 495 499 499 499 501 503 504 505 505 507 508 509 512 515
474
A.V. Banerjee and E. Duflo 4.5. The family: incomplete contracts within and across generations 4.6. Behavioral issues
5. Calibrating the impact of the misallocation of capital 5.1. A model with diminishing returns 5.2. A model with fixed costs 5.2.1. Taking stock
6. Towards a non-aggregative growth theory 6.1. An illustration 6.2. Can we take this model to the data? 6.2.1. What are the empirical implications of the above model? 6.2.2. Empirical evidence 6.3. Where do we go from here?
Acknowledgements References
518 520 522 523 527 534 535 535 538 538 541 542 544 544
Abstract Growth theory has traditionally assumed the existence of an aggregate production function, whose existence and properties are closely tied to the assumption of optimal resource allocation within each economy. We show extensive evidence, culled from the micro-development literature, demonstrating that the assumption of optimal resource allocation fails radically. The key fact is the enormous heterogeneity of rates of return to the same factor within a single economy, a heterogeneity that dwarfs the cross-country heterogeneity in the economy-wide average return. Prima facie, we argue, this evidence poses problems for old and new growth theories alike. We then review the literature on various causes of this misallocation. We go on to calibrate a simple model which explicitly introduces the possibility of misallocation into an otherwise standard growth model. We show that, in order to match the data, it is enough to have misallocated factors: there also needs to be important fixed costs in production. We conclude by outlining the contour of a possible non-aggregate growth theory, and review the existing attempts to take such a model to the data.
Keywords non-aggregative growth theory, aggregate production function, factor allocation, non-convexities JEL classification: O0, O10, O11, O12, O14, O15, O16, O40
Ch. 7: Growth Theory through the Lens of Development Economics
475
1. Introduction: neo-classical growth theory The premise of neo-classical growth theory is that it is possible to do a reasonable job of explaining the broad patterns of economic change across countries, by looking at it through the lens of an aggregate production function. The aggregate production function relates the total output of an economy (a country, for example) to the aggregate amounts of labor, human capital and physical capital in the economy, and some simple measure of the level of technology in the economy as a whole. It is formally represented as F(A, K P , K H , L) where K P and K H are the total amounts of physical and human capital invested, L is the total labor endowment of the economy and A is a technology parameter. The aggregate production function is not meant to be something that physically exists. Rather, it is a convenient construct. Growth theorists, like everyone else, have in mind a world where production functions are associated with people. To see how they proceed, let us start with a model where everyone has the option of starting a firm, and when they do, they have access to an individual production function Y = F (KP , KH , L, θ ),
(1)
where KP and KH are the amounts of physical and human capital invested in the firm and L is the amount of labor. θ is a productivity parameter which may vary over time, but at any point of time is a characteristic of the firm’s owner. Assume that F is increasing in all its inputs. To make life simpler, assume that there is only one final good in this economy and physical capital is made from it. Also assume that the population of the economy is described by a distribution function Gt (W, θ ), the joint distribution of W and θ , where W is the wealth of a particular individual and θ is his productivity ), the corresponding partial distribution on θ , be atomless. parameter. Let G(θ The lives of people, as is often the case in economic models, is rather dreary: In each period, each person, given his wealth, his θ and the prices of the inputs, decides whether to set up a firm, and if so how to invest in physical and human capital. At the end of the period, once he gets returns from the investment and possibly other incomes, he consumes and the period ends. The consumption decision is based on maximizing the following utility function: ∞
δ t U (Ct , θ ),
0 < δ < 1.
(2)
t=0
1.1. The aggregate production function The key assumption behind the construction of the aggregate production function is that all factor markets are perfect, in the sense that individuals can buy or sell as much as they want at a given price. With perfect factor markets (and no risk) the market must allocate the available supply of inputs to maximize total output. Assuming that the distribution of productivities does not vary across countries, we can therefore define F(K P , K H , L)
476
A.V. Banerjee and E. Duflo
to be:
max
{KP (θ),KH (θ),L(θ)}
θ
subject to KP (θ ) dθ = K P , θ
F KP (θ ), KH (θ ), L(θ ), θ dG(θ )
KH (θ ) dθ = K H ,
θ
L(θ ) dθ = L.
and θ
This is the aggregate production function. It is notable that the distribution of wealth does not enter anywhere in this calculation. This reflects the fact that with perfect factor markets, there is no necessary link between what someone owns and what gets ) does not enter as an argument of used in the firm that he owns. The fact that G(θ F(K P , K H , L) reflects our assumption that the distribution of productivities does not vary across countries. It should be clear from the construction that there is no reason to expect a close relation between the “shape” of the individual production function and the shape of the aggregate function. Indeed it is well known that aggregation tends to convexify the production set: In other words, the aggregate production function may be concave even if the individual production functions are not. In this environment where there are a continuum of firms, the (weak) concavity of the aggregate production function is guaranteed as long as the average product of the inputs in the individual production functions is bounded in the sense that there is a λ such that F (λKP , λKH , λL, θ ) λ(KP , KH , L, θ ) for all KP , KH , L and θ . It follows that the concavity of the individual functions is sufficient for the concavity of the aggregate but by no means necessary: The aggregate production function would also be concave if the individual production functions were S-shaped (convex to start out and then becoming concave). Alternately, the individual production function being bounded is enough to guarantee concavity of the aggregate production function. Moreover, the aggregate production function will typically be differentiable almost everywhere. It is a corollary of this result that the easiest way to generate an aggregate production function with increasing returns is to base the increasing returns not on the shape of the individual production function, but rather on the possibility of externalities across firms. If there are sufficiently strong positive externalities between investment in one firm and investment in another, increasing the total capital stock in all of them together will increase aggregate output by more (in proportional terms) than the same increase in a single firm would raise the firm’s output, which could easily make the aggregate production function convex. This is the reason why externalities have been intimately connected, in the growth literature, with the possibility of increasing returns. The assumption of perfect factor markets is therefore at the heart of neo-classical growth theory. It buys us two key properties: The fact that the ownership of factors does not matter, i.e., that an aggregate production function exists; and that it is concave. The next sub-section shows how powerful these two assumptions can be.
Ch. 7: Growth Theory through the Lens of Development Economics
477
1.2. The logic of convergence Assume for simplicity that production only requires physical capital and labor and that the aggregate production function, F(K p , L) as defined above, exhibits constant returns and is concave, increasing, almost everywhere differentiable and eventually strictly con . As noted above, this does not cave, in the sense that F < ε < 0, for any K p > K p require the individual production functions to have this shape, though it does impose some constraints on what the individual functions can be like. It does however require that the distribution of firm-level productivities is the same everywhere. Under our assumption that capital markets are perfect, in the sense that people can borrow and lend as much as they want at the common going rate, rt , the marginal returns to capital must be the same for everybody in the economy. This, combined with the preferences as represented by (2), has the immediate consequence that for everybody in the economy: U (Ct , θ) = δrt U (Ct+1 , θ). It follows that everybody’s consumption in the economy must grow as long as δrt > 1 and shrink if δrt < 1. And since consumption must increase with wealth, it follows that everyone must be getting richer if and only if δrt > 1, and consequently the aggregate wealth of the economy must be growing as long as δrt > 1. In a closed economy, the total wealth must be equal to the total capital stock, and therefore the capital stock must also be increasing under the same conditions. Credit market equilibrium, under perfect capital markets, implies that F (K P t , L) = rt . The fact that F is eventually strictly concave implies that as the aggregate capital stock grows, its marginal product must eventually start falling, at a rate bounded away from 0. This process can only stop when δF (K P t , L) = 1. As long as the production function is the same everywhere, all countries must end up equally wealthy. The logic of convergence starts with the fact that in poor countries capital is scarce, which combined with the concavity of the aggregate production function implies that the return on the capital stock should be high. Even with the same fraction of these higher returns being reinvested, the growth rate in the poorer countries would be higher. Moreover, the high returns should encourage a higher reinvestment rate, unless the income effect on consumption is strong enough to dominate. Together, they should make the poorer countries grow faster and catch up with the rich ones. Yet poorer countries do not grow faster. According to Mankiw, Romer and Weil (1992), the correlation between the growth rate and the initial level of Gross Domestic Product is small, and if anything, positive (the coefficient of the log of the GDP in 1960 on growth rate between 1960 and 1992 is 0.0943). Somewhere along the way, the logic seems to have broken down. Understanding the failure of convergence has been one of the key endeavors of the economics of growth. What we try to do in this chapter is to argue that the failure of this approach is intimately tied to the failure of the assumptions that underlie the
478
A.V. Banerjee and E. Duflo
construction of the aggregate production function and to suggest an alternative approach to growth theory that abandons the aggregate production. We start by discussing, in Section 2, the two implications of the neo-classical model that are at the root of the convergence result: Both rates of returns and investment rates should be higher in poor countries. We show that, in fact, neither rates of returns nor investment are, on average, much higher in poor countries. Moreover, contrary to what the aggregate production approach implies, there are large variations in rate of returns within countries, and large variation in the extent to which profitable investment opportunities are exploited. In Section 3, we ask whether the puzzle (of no convergence) can be solved, while maintaining the aggregate production function, by theories that focus on reasons for technological backwardness in poor countries. We argue that this class of explanations is not consistent with the empirical evidence which suggests that many firms in poor countries do use the latest technologies, while others in the same country use obsolete modes of production. In other words, what we need to explain is less the overall technological backwardness and more why some firms do not adopt profitable technologies that are available to them (though perhaps not affordable). In Section 4, we attempt to suggest some answers to the question of why firms and people in developing countries do not always avail themselves of the best opportunities afforded to them. We review various possible sources of the inefficient use of resources: government failures, credit constraints, insurance failure, externalities, family dynamics, and behavioral issues. We argue that each of these market imperfections can explain why investment may not always take place where the rates of returns are the highest, and therefore why resources may be misallocated within countries. This misallocation, in turn, drives down returns and this may lower the overall investment rate. In Section 5, we calibrate plausible magnitudes for the aggregate static impact of misallocation of capital within countries. We show that, combined with individual production functions characterized by fixed costs, the misallocation of capital implied by the variation of the returns to capital observed within countries can explain the main aggregate puzzles: the low aggregate productivity of capital, and the low Total Factor Productivity in developing countries, relative to rich countries. Non-aggregative growth models thus seem to have the potential to explain why poor countries remain poor. The last section provides an introduction to an alternative growth theory that does not require the existence of an aggregate production function, and therefore can accommodate the misallocation of resources. We then review the attempts to empirically test these models. We argue that the failure to take seriously the implications of non-aggregative models have led to results that are very hard to interpret. To end, we discuss an alternative empirical approach illustrated by some recent calibration exercises based on growth models that take the misallocation of resources seriously.
Ch. 7: Growth Theory through the Lens of Development Economics
479
2. Rates of return and investment rates in poor countries In this section, we examine whether the two main implications of the neo-classical model are verified in the data: Are returns and investment rates higher in poor countries? 2.1. Are returns higher in poor countries? 2.1.1. Physical capital • Indirect estimates One way to look at this question is to look at the interest rates people are willing to pay. Unless people have absolutely no assets that they can currently sell, the marginal product of whatever they are doing with the marginal unit of capital should be no less than the interest rate: If this were not true, they could simply divert the last unit of capital toward whatever they are borrowing the money for and be better off. There is a long line of papers that describe the workings of credit markets in poor countries [Banerjee (2003) summarizes this evidence]. The evidence suggests that a substantial fraction of borrowing takes place at very high interest rates. A first source of evidence is the “Summary Report on Informal Credit Markets in India” [Dasgupta (1989)], which reports results from a number of case studies that were commissioned by the Asian Development Bank and carried out under the aegis of the National Institute of Public Finance and Policy. For the rural sector, the data is based on surveys of six villages in Kerala and Tamil Nadu, carried out by the Centre for Development Studies. The average annual interest rate charged by professional moneylenders (who provide 45.6% of the credit) in these surveys is about 52%. For the urban sector, the data is based on various case surveys of specific classes of informal lenders, many of whom lend mostly to trade or industry. For finance corporations, they report that the minimum lending rate on loans of less than one year is 48%. For hire-purchase companies in Delhi, the lending rate was between 28% and 41%. For auto financiers in Namakkal, the lending rate was 40%. For handloom financiers in Bangalore and Karur, the lending rate varied between 44% and 68%. Several other studies reach similar conclusions. A study by Timberg and Aiyar (1984) reports data on indigenous-style bankers in India, based on surveys they carried out: The rates for Shikarpuri financiers varied between 21% and 37% on loans to members of local Shikarpuri associations and between 21% and 120% on loans to non-members (25% of the loans were to non-members). Aleem (1990) reports data from a study of professional moneylenders that he carried out in a semi-urban setting in Pakistan in 1980–1981. The average interest rate charged by these lenders is 78.5%. Ghate (1992) reports on a number of case studies from all over Asia: The case study from Thailand found that interest rates were 5–7% per month in the north and northeast (5% per month is 80% per year and 7% per month is 125%). Murshid (1992) studies Dhaner Upore (cash for kind) loans in Bangladesh (you get some amount in rice now and repay some amount in rice later) and reports that the interest rate is 40% for a 3–5 month loan
480
A.V. Banerjee and E. Duflo
period. The Fafchamps (2000) study of informal trade credit in Kenya and Zimbabwe reports an average monthly interest rate of 2.5% (corresponding to an annualized rate of 34%) but also notes that this is the rate for the dominant trading group (Indians in Kenya, whites in Zimbabwe), while the blacks pay 5% per month in both places. The fact that interest rates are so high could reflect the high risk of default. However, this does not appear to be the case, since several of studies mentioned above give the default rates that go with these high interest rates. The study by Dasgupta (1989) attempts to decompose the observed interest rates into their various components,1 and finds that the default costs explain 7 per cent (not 7 percentage points!) of the total interest costs for auto financiers in Namakkal and handloom financiers in Bangalore and Karur, 4% for finance companies and 3% for hire-purchase companies. The same study reports that in four case studies of moneylenders in rural India they found default rates explained about 23% of the observed interest rate. Timberg and Aiyar (1984), whose study is also mentioned above, report that average default losses for the informal lenders they studied ranges between 0.5% and 1.5% of working funds. The study by Aleem (1990) gives default rates for each individual lender. The median default rate is between 1.5 and 2%, and the maximum is 10%.2 Finally, it does not seem to be the case that these high rates are only paid by those who have absolutely no assets left. The “Summary Report on Informal Credit Markets in India” [Dasgupta (1989)] reports that several of the categories of lenders that have already been mentioned, such as handloom financiers and finance corporations, focus almost exclusively on financing trade and industry while Timberg and Aiyar (1984) report that for Shikarpuri bankers at least 75% of the money goes to finance trade and, to lesser extent, industry. In other words, they only lend to established firms. It is hard to imagine, though not impossible, that all the firms have literally no assets that they can sell. Ghate (1992) also concludes that the bulk of informal credit goes to finance trade and production, and Murshid (1992), also mentioned above, argues that most loans in his sample are production loans despite the fact that the interest rate is 40% for a 3–5 month loan period. Udry (2003) obtains similar indirect estimates by restricting himself to a sector where loans are used for productive purpose, the market for spare taxi parts in Accra, Ghana. He collected 40 pairs of observations on price and expected life for a particular used car part sold by a particular dealer (e.g., alternator, steering rack, drive shaft). Solving for the discount rate which makes the expected discounted cost of two similar parts equal gives a lower bound to the returns to capital. He obtains an estimate of 77% for the median discount rate. 1 In the tradition of Bottomley (1963). 2 Here we make no attempt to answer the question of why the interest rates are so high. Banerjee (2003)
argues that it is not implausible that the enormous gap between borrowing and lending rates implied by these numbers simply reflects the cost of lending (monitoring and contracting costs of various kinds). Hoff and Stiglitz (1998) suggest an important role for monopolistic competition, in the presence of a fixed cost of lending. There is also a view that the market for credit is monopolized by a small number of lenders who earn excess profits, but Aleem (1990) finds no evidence of excess profits.
Ch. 7: Growth Theory through the Lens of Development Economics
481
Together, these studies thus suggest that people are willing to pay high interest rates for loans used for productive purpose, which suggests that the rates of return to capital are indeed high in developing countries, at least for some people. • Direct estimates Some studies have tried to come up with more direct estimates of the rates of returns to capital. The “standard” way to estimate returns to capital is to posit a production function (translog and Cobb–Douglas, generally) and to estimate its parameters using OLS regression, or instrumenting capital with its price. Using this methodology, Bigsten et al. (2000) estimate returns to physical and human capital in five African countries. They estimate rates of returns ranging from 10% to 32%. McKenzie and Woodruff (2003) estimate parametric and non-parametric relationships between firm earnings and firm capital. Their estimates suggest huge returns to capital for these small firms: For firms with less than $200 invested, the rate of returns reaches 15% per month, well above the informal interest rates available in pawn shops or through micro-credit programs (on the order of 3% per month). Estimated rates of return decline with investment, but remain high (7% to 10% for firms with investment between $200 and $500, 5% for firms with investment between $500 and $1,000). Such studies present serious methodological issues, however. First, the investment levels are likely to be correlated with omitted variables. For example, in a world without credit constraints, investment will be positively correlated with the expected returns to investment, generating a positive “ability bias” [Olley and Pakes (1996)]. McKenzie and Woodruff attempt to control for managerial ability by including the firm owner’s wage in previous employment, but this may go only part of the way if individuals choose to enter self-employment precisely because their expected productivity in self-employment is much larger than their productivity in an employed job. Conversely, there could be a negative ability bias, if capital is allocated to firms in order to avoid their failure. Banerjee and Duflo (2004) take advantage of a change in the definition of the socalled “priority sector” in India to circumvent these difficulties. All banks in India are required to lend at least 40% of their net credit to the “priority sector”, which includes small-scale industry, at an interest rate that is required to be no more than 4% above their prime lending rate. In January, 1998, the limit on total investment in plants and machinery for a firm to be eligible for inclusion in the small-scale industry category was raised from Rs. 6.5 million to Rs. 30 million. In 2000, the limit was lowered back to Rs. 10 million. Banerjee and Duflo (2004) first show that, after the reforms, newly eligible firms (those with investment between 6.5 million and 30 million) received on average larger increments in their working capital limit than smaller firms. They then show that the sales and profits increased faster for these firms during the same period. The opposite happened when the priority sector was contracted again. Putting these two facts together, they use the variation in the eligibility rule over time to construct instrumental variable estimates of the impact of working capital on sales and profits. After computing a non-subsidized cost of capital, they estimate that the returns to capital in these firms must be at least 74%.
482
A.V. Banerjee and E. Duflo
There is also direct evidence of very high rates of returns on productive investment in agriculture. Goldstein and Udry (1999) estimate the rates of returns to the production of pineapple in Ghana. The rate of returns associated with switching from the traditional maize and Cassava intercrops to pineapple is estimated to be in excess of 1200%! Few people grow pineapple, however, and this figure may hide some heterogeneity between those who have switched to pineapple and those who have not. Evidence from experimental farms also suggests that, in Africa, the rate of returns to using chemical fertilizer (for maize) would also be high. However, this evidence may not be realistic, if the ideal conditions of an experimental farm cannot be reproduced on actual farms. Foster and Rosenzweig (1995) show, for example, that the returns to switching to high yielding varieties were actually low in the early years of the green revolution in India, and even negative for farmers without an education. This is despite the fact that these varieties had precisely been selected for having high yields, in proper conditions. But they required complementary inputs in the correct quantities and timing. If farmers were not able or did not know how to supply those, the rates of returns were actually low. To estimate the rates of returns to using fertilizer in actual farms in Kenya, Duflo, Kremer and Robinson (2003), in collaboration with a small NGO, set up small scale randomized trials on people’s farms: Each farmer in the trials designated two small plots. On one randomly selected plot, a field officer from the NGO helped the farmer apply fertilizer. Other than that, the farmers continued to farm as usual. They find that the rates of returns from using a small amount of fertilizer varied from 169% to 500% depending on the year, although of returns decline fast with the quantity used on a plot of a given size. This is not inconsistent with the results in Foster and Rosenzweig (1995), since by the time this study was conducted in Kenya, chemical fertilizer was a well established and well understood technology, which did not need many complementary inputs. The direct estimates thus tend to confirm the indirect estimates: While there are some settings where investment is not productive, there seems to be investment opportunities which yield substantial rates of returns. • How high is the marginal product on average? The fact that the marginal product in some firms is 50% or 100% or even more does not imply that the average of the marginal products across all firms is nearly as high. Of course, if capital always went to its best use, the notion of the average of the marginal products does not make sense. The presumption here is that there may be an equilibrium where the marginal products are not equalized across firms. One way to get at the average of the marginal products is to look at the Incremental Capital–Output Ratio (ICOR) for the country as a whole. The ICOR measures the increase in output predicted by a one unit increase in capital stock. It is calculated by extrapolating from the past experience of the country and assumes that the next unit of capital will be used exactly as efficiently (or inefficiently) as the last one. The inverse of the ICOR therefore gives an upper bound for the average marginal product for the economy – it is an upper bound because the calculation of the ICOR does not control
Ch. 7: Growth Theory through the Lens of Development Economics
483
for the effect of the increases in the other factors of production which also contributes to the increase in output.3 For the late 1990s, the IMF estimates that the ICOR is over 4.5 for India and 3.7 for Uganda. The implied upper bound on the average marginal product is 22% for India and 27% in Uganda. This is also consistent with the work of Pessoa, Cavalcanti-Ferreira and Velloso (2004) who estimate a production function using crosscountry data and calculate marginal products for developing countries which are in the 10–20% range. It seems that the average returns are actually not much higher than 9% or so, which is the usual estimate for the average stock market return in the U.S. • Variations in the marginal products across firms To reconcile the high direct and indirect estimates of the marginal returns we just discussed and an average marginal product of 22% in India, it would have to be that there is substantial variation in the marginal product of capital within the country. Given that the inefficiency of the Indian public sector is legendary, this may just be explained by the investment in the public sector. However, since the ICOR is from the late 1990s, when there was little new investment (or even disinvestment) in the public sector, there must also be many firms in the private sector with marginal returns substantially below 22%. The micro evidence reported in Banerjee (2003), which shows that there is very substantial variation in the interest rate within the same sub-economy, certainly goes in this direction. The Timberg and Aiyar (1984) study mentioned above, is one source of this evidence: It reports that the Shikarpuri lenders charged rates that were as low as 21% and as high as 120%, and some established traders on the Calcutta and Bombay commodity markets could raise funds for as little as 9%. The study by Aleem (1990), also mentioned above, reports that the standard deviation of the interest rate was 38.14%. Given that the average lending rate was 78.5%, this tells us that an interest rate of 2% and an interest rate of 150% were both within two standard deviations of the mean. Unfortunately, we cannot quite assume from this that there are some borrowers whose marginal product is 9% or less: The interest rate may not be the marginal product if the borrowers who have access to these rates are credit constrained. Nevertheless, given that these are typically very established traders, this is less likely than it would be otherwise. Ideally we would settle this issue on the basis of direct evidence on the misallocation of capital, by providing direct evidence on variations in rates of return across groups of firms. Unfortunately such evidence is not easy to come by, since it is difficult to consistently measure the marginal product of capital. However, there is some rather suggestive evidence from the knitted garment industry in the Southern Indian town of Tirupur [Banerjee and Munshi (2004), Banerjee, Duflo and Munshi (2003)]. Two groups of people operate in Tirupur: the Gounders, who issue from a small, wealthy, agricultural community from the area around Tirupur, who have moved into the ready-made garment industry because there was not much investment opportunity in agriculture. Outsiders from various regions and communities started joining the city in the 1990s. 3 The implicit assumption that the other factors of production are growing is probably reasonable for most developing countries, except perhaps in Africa.
484
A.V. Banerjee and E. Duflo
The Gounders have, unsurprisingly, much stronger ties in the local community, and thus better access to local finance, but may be expected to have less natural ability for garment manufacturing than the outsiders, who came to Tirupur precisely because of its reputation as a center for garment export. The Gounders own about twice as much capital as the outsiders on average. They maintain a higher capital–output ratio than the outsiders at all levels of experience, though the gap narrows over time. The data also suggest that they make less good use of their capital than the outsiders: While the outsiders start with lower production and exports than the Gounders, their experience profile is much steeper, and they eventually overtake the Gounders at high levels of experience, even though they have lower capital stock throughout. This data therefore suggests that capital does not flow where the rates of return are highest: The outsiders are clearly more able than the Gounders, but they nevertheless invest less.4 To summarize, the evidence on returns to physical capital in developing countries suggests that there are instances with high rates of return, while the average of the marginal rates of return across firms does not appear to be that high. This suggests a coexistence of very high and very low rates of return in the same economy. 2.1.2. Human capital • Education The standard source of data on the rate of return to education is Psacharopoulos and Patrinos (1973, 1985, 1994, 2002) who compiles average Mincerian returns to education (the coefficient of years of schooling in a regression of log(wages) on years of schooling) as well as what he call “full returns” to education by level of schooling. Compared to Mincerian returns, full returns take into account the variation in the cost of schooling according to year of schooling: The opportunity cost of attending primary school is low, because 6 to 12-year-old children do not earn the same wage as adults; and the direct costs of education increase with the level of schooling. On the basis of this data, Psacharopoulos argues that returns to education are substantial, and that they are larger in poor countries than in rich countries. We re-examine the claim that returns to education are larger in poor countries, using data on traditional Mincerian returns, which have the advantage of being directly comparable. We start with the latest compilation of rates of returns, available in Psacharopoulos and Patrinos (2002) and on the World Bank web site. We update it as much as possible, using studies that seem to have been overlooked by Psacharopoulos, or that have appeared since then (the updated data set and the references are presented in Table 1).5 We flag the observations that Bennell (1996) rated as being of “poor” or “very poor” quality. We complete 4 This is not because capital and talent happen to be substitutes. In this data, as it is generally assumed,
capital and ability appear to be complements. 5 The bulk of the update is for African countries, where Bennell (1996) had systematically investigated the Psacharoupoulos data, and found that many of the underlying studies were unreliable.
Country
Year
Mincerian returns
Years of schooling (Psacharopoulos)
Years of schooling (World Bank)
Source
South America Australia Europe South America Africa South America Africa Africa North America South America Asia South America South America Africa Europe Europe South America South America Africa South America Europe Africa Europe Europe Europe Africa Europe
1989 1989 1993 1993 1979 1998 1980 1995 1989 1989 1993 1989 1992 1987 1994 1990 1989 1987 1997 1992 1994 1997 1993 1977 1988 1999 1993
10.3 8 7.2 10.7 19.1 12.21 9.6 5.96 8.9 12 12.2 14 8.50 13.10 5.2 4.5 9.4 11.8 7.80 7.6 5.4 3.28 8.2 10 7.7 8.80 7.6
9.1
8.83 10.92 8.35 5.58 6.28 4.88
Psacharopoulos (1994) Cohn and Addison (1998) Fersterer and Winter-Ebmer (1999) Patrinos (1995) Psacharopoulos (1994) Verner (2001) Psacharopoulos (1994) Appleton et al. (1999) Cohn (1997) Psacharopoulos (1994) Hossain (1997) Psacharopoulos (1994) Funkhouser (1998) Schultz (1994) Menon (1995) Christensen and Westergard-Nielsen (1999) Psacharopoulos (1994) Psacharopoulos (1994) Wahba (2000) Funkhouser (1996) Kroncke (1999) Krishnan, Selasie, Dercon (1989) Asplund (1999) Psacharopoulos (1994) Cohn and Addison (1998) Frazer (1998) Magoula and Psacharopoulos (1999)
3.3 5.3
8.5 8.2
3.54 11.62 7.55 6.36 5.27 6.05
6.9
8.8 9.6
9.15 9.66 4.93 6.41 5.51 5.15
10.9 6 6.2 9.7
9.99 7.86 10.2 3.89 8.67
Data rating (Bennel)
Additions to Psacharopoulos data
Poor Added Poor Added
Poor
Added Added
Poor
Added
Added 485
Argentina Australia Austria Bolivia Botswana Brazil Burkina Faso Cameroon Canada Chile China Colombia Costa Rica Cote d’lvoire Cyprus Denmark Dominican Rep. Ecuador Egypt El Salvador Estonia Ethiopia Finland France Germany Ghana Greece
Continent
Ch. 7: Growth Theory through the Lens of Development Economics
Table 1 Rate of returns to education and years of schooling
486
Table 1 (Continued) Country
Year
Mincerian returns
Years of schooling (Psacharopoulos)
Years of schooling (World Bank)
Source
South America South America Asia Europe Asia Asia Asia Asia Europe South America Asia Africa Asia Asia Asia South America Africa Asia Europe South America Europe Asia South America South America South America South America
1989 1991 1981 1987 1995 1995 1975 1979 1987 1989 1988 1995 1986 1983 1979 1997 1970 1999 1994 1996 1995 1991 1990 1990 1990 1998
14.9 9.3 6.1 4.3 10.6 7 11.6 6.4 2.7 28.8 13.2 11.39 13.5 4.5 9.4 35.31 15.8 9.7 6.4 12.1 5.5 15.4 13.7 11.5 8.1 12.6
4.3
3.49
9.1 11.3
4.8 9.13 5.06 4.99 5.31 9.6 7.18 5.26 9.47 4.2 10.84 7.05 6.8 7.23
Psacharopoulos (1994) Funkhouser (1996) Psacharopoulos (1994) Psacharopoulos (1994) Kingdon (1998) Duflo (2000) Psacharopoulos (1994) Psacharopoulos (1994) Brunello, Comi and Lucifora (1999) Psacharopoulos (1994) Cohn and Addison (1998) Appleton et al. (1998) Ryoo, Nam and Carnoy (1993) Psacharopoulos (1994) Psacharopoulos (1994) Lopez-Acevedo (2001) Psacharopoulos (1994) Parajuli (1999) Hartog, Odink and Smits (1999) Belli and Ayadi (1998) Earth and Roed (1999) Katsis, Mattson and Psacharopoulos (1998) Psacharopoulos (1994) Psacharopoulos (1994) Psacharopoulos (1994) Schady (2000)
8 11.2 7.2 8 8 8.9 15.8 2.9 3.9
9.2 9.1 10.1 8.8
2.43 9.36 4.58 11.85 3.88 8.55 6.18 7.58 8.21
Data rating (Bennel)
Additions to Psacharopoulos data
Poor Poor Poor Added Poor Added Poor A.V. Banerjee and E. Duflo
Guatemala Honduras Hong Kong Hungary India Indonesia Iran Israel Italy Jamaica Japan Kenya Korea Kuwait Malaysia Mexico Morocco Nepal Netherlands Nicaragua Norway Pakistan Panama Paraguay Peru Philippines
Continent
Country
Poland Portugal Puerto Rico Russian Federation Singapore South Africa Spain Sri Lanka Sudan Sweden Switzerland Taiwan Tanzania Thailand Tunisia Uganda United Kingdom United States Uruguay Venezuela Vietnam Yugoslavia Zambia Zimbabwe
Continent
Year
Mincerian returns
Europe Europe South America Europe Asia Africa Europe Asia Africa Europe Europe Asia Africa Asia Africa Africa Europe North America South America South America Asia Europe Africa Africa
1996 1991 1989 1996 1998 1993 1991 1981 1989 1991 1991 1998 1991 1989 1980 1992 1987 1995 1989 1992 1992 1986 1995 1994
7 8.6 15.1 7.2 13.1 10.27 7.2 7 9.3 5 7.5 19.01 13.84 11.5 8 5.94 6.8 10 9.7 9.4 4.8 4.8 10.65 5.57
Years of schooling (Psacharopoulos)
11.7 9.5 7.1 4.5 10.2
Years of schooling (World Bank)
Source
9.84 5.87
Nesterova and Sabirianova (1998) Cohn and Addison (1998) Griffin and Cox Edwards (1993) Nesterova and Sabirianova (1998) Sakellariou (2001) Mwabu and Schultz (1995) Mora (1999) Psacharopoulos (1994) Cohen and House (1994) Cohn and Addison (1998) Weber and Wolter (1999) Vere (2001) Mason and Kandker (1995) Patrinos (1995) Psacharopoulos (1994) Appleton et al. (1996) Psacharopoulos (1994) Rouse (1999) Psacharopoulos (1994) Psacharopoulos and Mattson (1998) Moock, Patrinos and Venkataraman (1998) Bevc (1993) Appleton et al. (1999) Appleton et al. (1999)
7.05 6.14 7.28 6.87 2.14 11.41 10.48
9
4.8 11.8 9
2.71 6.5 5.02 3.51 9.42 12.05 7.56 6.64
7.9 5.46 5.35
Data rating (Bennel)
Additions to Psacharopoulos data
Added
Poor
Added Added
Poor
Ch. 7: Growth Theory through the Lens of Development Economics
Table 1 (Continued)
Added
Added Added
487
Notes: This table updates Psacharopoulos and Patrinos (2002). The last column indicates which rate of returns were added by us. The data rating quality is from Bennell (1996), and concerns only African Countries.
488
A.V. Banerjee and E. Duflo
this updated database by adding data on years of schooling for the year of the study when it was not reported by Psacharopoulos. Using the preferred data, the Mincerian rates of returns seem to vary little across countries: The mean rate of returns is 8.96, with a standard deviation of 2.2. The maximum rate of returns to education (Pakistan) is 15.4%, and the minimum is 2.7% (Italy). Averaging within continents, the average returns are highest in Latin America (11%) and lowest in the Europe and the U.S. (7%), with Africa and Asia in the middle. If we run an OLS regression of the rates of returns to education on the average educational attainment (number of years of education), using the preferred data (updated database without the low quality data), the coefficient is −0.26, and is significant at 10% level (Table 2, column (3)). The returns to education predicted from this regression range from 6.9% for the country with the lowest education level to 10.1% for the country with the highest education level. This is a small range (smaller than the variation in the estimates of the returns to education of a single country, or even in different specifications in a single paper!): There is therefore no prima facie evidence that returns to education are much higher when education is lower, although the relationship is indeed negative. Columns (1) and (2) in the same table show that the data construction matters: When the countries with “poor” quality are included, the coefficient of years of education increases to −0.45. When only the 38 countries in the latest Psacharopoulos update are included (most countries are dropped because the database does not report years of education, even for countries where it is clearly available – Austria for example), the coefficient more than doubles, to −0.71. On the whole, this strong negative number does appear to be an artifact of data quality. In column (4), we directly regress the Mincerian returns to education on GDP, and we find a small and significant negative relationship. However, this is counteracted by the fact that teacher salary grows less fast than GDP, and the cost of education is thus not proportional to GDP: In column (5) we regress the log of the teacher salary on the log of GDP per capita.6 The coefficient is significantly less than one, suggesting that teachers are relatively more expensive in poor countries. This is to some extent attenuated by the fact that class sizes are larger in poor countries (which tends to make education cheaper). We then compute the returns to educating a child for one year as the ratio of the lifetime benefit of one year of education (assuming a life span of 30 years, a discount rate of 5%, a share of wage in GDP of 60%, and no growth), to the direct cost of education (assuming that teacher salary is 85% of the cost of education). In column (6), we regress this ratio on GDP: There is no relationship between this measure of returns and GDP.7 If we factor in indirect costs (as a fraction of GDP) (in column (7)), the relationship becomes slightly more negative, but still insignificant. On balance, the returns to one more year of education are therefore no higher in poor countries. 6 The teacher salary data is obtained from the “Occupational Wages Around the World” database [Freeman
and Oostendorp (2001)]. 7 Note that by assuming that the lifespan is the same in poor and rich countries, we are biasing upwards the returns in poor countries.
Variable Sample
Constant Mean years of schooling
Mincerian returns Psacharopoulos
direct costs/benefits
total costs/benefits
Psacharopoulos high quality (3)
Psacharopoulos high quality (4)
(5)
(6)
(7)
(1)
Psacharopoulos extended (2)
16.40 (2.6) −0.72 (0.3)
13.01 (1.35) −0.47 (0.16)
11.04 (1.14) −0.27 (0.14)
9.65 (0.46)
2.24 (0.15)
4.09 (0.21)
21.43 (1.63)
−0.034 (0.019)
−0.155 (0.147)
61 0.05
61 0.018
−0.084 (0.039)
GDP/capita (*1000) lgdp n r2
log(teacher salary)
37 0.139
70 0.106
62 0.062
62 0.072
0.79 (0.02) 532 0.7902
Ch. 7: Growth Theory through the Lens of Development Economics
Table 2 Returns to education
Source: The data on returns to education was compiled starting from Psacharopoulos and Patrinos (2002) and extended by surveying the literature. Table 1 lists the data and the sources. The data on teacher salary is from Freeman and Oosterkberke. The data on pupil teacher ratio is from the UNESCO Institute for Statistics database, available at http://www.uis.unesco.org/.
489
490
A.V. Banerjee and E. Duflo
• Health Education is not the only dimension of human capital. In developing countries, investment in nutrition and health has been hypothesized to have potentially high returns at moderate levels of investment. The report of the Commission for Macroeconomics and Health [Commission on Macroeconomics and Health (2001)], for example, estimated returns to investing in health to be on the order of 500%, mostly on the basis of cross-country growth regressions. Several excellent recent surveys [Strauss and Thomas (1995, 1998), Thomas (2001) and Thomas and Frankenberg (2002)] summarize the existing literature on the impact of different measures of health on fitness and productivity, and lead to a much more nuanced conclusion. There is substantial experimental evidence that supplementation in iron and vitamin A increases productivity at relatively low cost. Unfortunately, not all studies report explicit rates of returns calculations. The few numbers that are available suggest that some basic health intervention can have high rates of returns: Basta et al. (1979) studies an iron supplementation experiment conducted among rubber tree tappers in Indonesia. Baseline health measures indicated that 45% of the study population was anemic. The intervention combined an iron supplement and an incentive (given to both treatment and control groups) to take the pill on time. Work productivity in the treatment group increased by 20% (or $132 per year), at a cost per worker-year of $0.50. Even taking into account the cost of the incentive ($11 per year), the intervention suggests extremely high rates of returns. Thomas et al. (2003) obtain lower, but still high, estimates in a larger experiment, also conducted in Indonesia: They found that iron supplementation experiments in Indonesia reduced anemia, increased the probability of participating in the labor market, and increased earnings of self-employed workers. They estimate that, for self-employed males, the benefits of iron supplementation amount to $40 per year, at a cost of $6 per year.8 The cost benefit analysis of a de-worming program [Miguel and Kremer (2004)] in Kenya reports estimates of a similar order of magnitude: Taking into account externalities (due to the contagious nature of worms), the program led to an average increase in school participation of 0.14 years. Using a reasonable figure for the returns to a year of education, this additional schooling will lead to a benefit of $30 over the life of the child, at a cost of $0.49 per child per year. Not all interventions have the same rates of return however: A study of Chinese cotton mill workers [Li et al. (1994)] led to a significant increase in fitness, but no corresponding increase in productivity. Likewise, the intervention analyzed by Thomas et al. (2003) had no effect on earnings or labor force participation of women. In summary, while there is not much debate on the impact of fighting anemia (through iron supplementation or de-worming) on work capacity, there is more heterogeneity amongst estimates of economic rates of return of these interventions. The heterogeneity is even larger when we consider other forms of health interventions, reviewed, for 8 This number takes into account the fact that only 20% of the Indonesian population is iron deficient: The private returns of iron supplementation for someone who knew they were iron deficient – which they can find out using a simple finger prick test – would be $200.
Ch. 7: Growth Theory through the Lens of Development Economics
491
example, in Strauss and Thomas (1995), or when one compares various human capital interventions. As in the case of physical capital, there are instances of high returns, and substantial heterogeneity in returns. 2.1.3. Taking stock: returns on capital The marginal product of physical and human capital in developing countries seems very high in some instances, but not necessarily uniformly. The average of the marginal products of physical capital in India may be as low as 22%, though even reasonably large firms often have marginal products of 60%, or even 100%. As long as we remain in the world of aggregative growth theory, the average marginal product is of course equal to the marginal product, since marginal products are always equated. Moreover even if there is some transitory variation in the marginal product, the relevant number from the point of view of any investor, should be the maximum and not the average: Capital should flow to where the returns are highest. The investments with returns of 60% or more should be the ones that guide investment, and not the 22%, and this ought to favor convergence. That being said, there is nothing in what we have said that tells us whether 22% is lower than what we would have predicted based on an aggregative growth model that predicts convergence, or is exactly right. Lucas (1990), in a well-known paper, suggests an approach to this question. He starts with the observation that according to the Penn World Tables [Heston, Summers and Aten (2002)], in 1990, output-per-worker in India at Purchasing Power Parity was 1/11th of what it was in the U.S. To obtain a productivity gap per effective use of labor, we need to adjust this ratio by the differences in education between the two countries. Based on the work of Krueger (1967), Lucas (1990) argues that “one American worker is equal to five Indian workers” in terms of human capital. In our case, since we are comparing productivity in 1990, and Krueger’s estimates of human capital are from the late 1960s, we presumably adjust the correction factor. Between 1965 and 1990, years of schooling among those 25 years or older went from 1.90 years to 3.68 years in India and from 9.25 years to 12 years in the United States, i.e., from approximately 20% of the U.S. level, which fits with the 5 : 1 gap in productivity that Krueger suggested, to about 30%.9 To show what this implies, Lucas starts with the assumption that net output is produced using a production function Y = AL1−α K α , where K is investment and L is the number of workers.10 9 These numbers are based on Barro and Lee (2000). Another angle from which this can be looked at is that health improved also during the period: Over a slightly different period, (1970–1975 to 1995–2000), according to the Human Development Report [United Nations Development Program (2001)], life expectancy at birth went from 50.3 to 62.3 years in India and from 71.5 years to 76.5 years in the U.S., reducing the gap between India and the U.S. by about 40%. 10 Lucas actually computes the ratio of output per effective unit of labor, which, with our parameters, is equal 3 ≈ 3. Reassuringly, this is also the ratio that Lucas started with, albeit based on the average numbers to 11 · 10 for the 1965–1990 period rather than the 1990 numbers.
492
A.V. Banerjee and E. Duflo
From this, it follows that output per worker is y = Ak α , where k is investment per worker in equipment. Assuming that firms can borrow as much as they want at the rate r, profit maximization requires that αAk α−1 = r, from which it follows that 1 α rI 1−α AU 1−α yU = . (3) yI rU AI If we assume that the only difference between the TFP levels in the two countries is due to the productivity per worker, the fact that Indian workers are only 30% as productive as the U.S. workers and the share of capital is assumed to be 40% implies that: AU = (0.3)−0.6 ≈ 2. (4) AI With these parameters, the 11-fold difference between yU and yI would imply that rI = (3.3)3/2 rU ≈ 6rU . r is naturally thought of as the marginal product of capital. In other words, if we take 9% for the marginal product of capital in the U.S., this would imply a 54% rate for India. Lucas, at this point, did not even wait to look at the data: If the difference in the returns were indeed so large, all the capital would flow from the U.S. to India. Hence, Lucas argued, the rate of returns cannot possibly be that high in India. As we know, this is something of a leap of faith, since capital does not flow even when there are large differences in returns within the same country. On the other hand, our estimates of the average marginal product is 22%. So Lucas was right in insisting that the actual rates of returns are much lower than what we would expect if the model were correct. This is strictly only true if we estimate the marginal product from the data on output per worker; however if we calculate it directly from the capital–labor ratio, the problem shows up elsewhere. To see this, recall from Equation (4) that assuming that workers are only 30% as productive is equivalent to assuming that TFP in India should be approximately 50% of what it is in the U.S. This, combined with the fact that, according to the Penn World Tables, the U.S. has 18 times more capital-per-worker than India implies that the marginal product of capital ought to be 12 (18)0.6 = 2.8 times higher in India, which tells us that the marginal product in India ought to be about 25%, which is probably close to what it is. However if we now put in the numbers for capital-per-worker into the production function, the ratio of output per worker in the two countries turns out to be: 0.4 kU yu =2 = 2 · (18)0.4 = 6.35. (5) yI kI In the data this ratio is 11 : 1. In other words, the problem is still there: Earlier when we used the capital–labor ratio implied by the low level of worker productivity it told us that the return on capital should be much higher than it is. On the other hand, when we use the actual capital–labor ratio, we see that the implied return on capital is quite
Ch. 7: Growth Theory through the Lens of Development Economics
493
reasonable, but the predicted worker productivity is much higher than it is in the data. Either way, it seems clear that we need to go beyond this model. 2.2. Investment rates in poor countries 2.2.1. Is investment higher in poor countries? Prima facie, it does not seem to be the case that investment rates are higher in poor countries. On the contrary, there is a robust positive correlation between investment rates in physical capital and income per capita, when both are expressed in terms of purchasing power parity. In fact, Levine and Renelt (1992) and Sala-i-Martin (1997) identified investment per capita as the only robust correlate of income. For example, Hsieh and Klenow (2003) estimate that in 1985, the correlation between PPP investment rate and PPP income per capita for the 115 countries present in the Penn World Tables was 0.60. The coefficients they estimate suggest that an increase in one log point in income per capita is associated with about a 5 percentage point higher PPP investment rate (the mean investment rate is 14.5%). The same positive correlation obtains with investment in plant and machinery. The relationship between investment rate and income per capita is much less strong when both of them are expressed in nominal terms rather than in PPP terms [Eaton and Kortum (2001), Restuccia and Urrutia (2001) and Hsieh and Klenow (2003)]. The coefficient drops by a third when all investments are considered, and becomes insignificant when the measure of investment includes only plant and machinery. According to Hsieh and Klenow (2003), the fact that poor countries have a lower investment-to-GDP ratio, when expressed in PPP, is explained by the low relative price of consumption, relative to investment: While there is no correlation between investment prices and GDP, there is a strong positive correlation between consumption prices and GDP. It is not clear, however, that knowing this helps us explain why there is not more investment in poor countries. First, because the high rates that we found in some firms in developing countries and the lower, but still much higher than U.S., rates that we found on average are there despite the high price of capital goods. This, by itself, should encourage investment, unless income effects are unusually strong. Moreover, even if we measure everything in nominal terms, there is no strong negative correlation between investment and GDP. There are, of course, examples of poor countries with large investment-to-GDP ratios. Young (1995) shows that a substantial fraction of the rapid growth of the East-Asian economies in the post-WWII period can be accounted for by rapid factor accumulation (including increase in the size of the labor force, factor reallocation, and high investment rates). In particular, according to the national accounts, between 1960 and 1985, the capital stock in Singapore, Korea, and Taiwan grew at more than 12% a year (in Hong Kong, it grew only at 7.7% a year). Between 1966 and 1999, the capital–output ratio has increased at an average rate of 3.4% a year in Korea, and 2.8% in Singapore. In Singapore, for example, the constant investment-to-GDP ratio increased from 10% in 1960 to 47% in 1984. In Singapore, Korea, and Taiwan, this increase in the
494
A.V. Banerjee and E. Duflo
stock of capital alone is responsible for about 1% out of the average yearly 3.4% to 4% of the “naive” Solow residual. Based on these results, Young (1995) concluded that the East-Asian economies are perfect examples of transitional dynamics in the neo-classical model. However, in subsequent research, Hsieh (1999) questioned the validity of the national account data for investment for Singapore. He observes that if the capital-to-GDP ratio had grown at that speed, one would have observed a commensurate reduction in the rental price of capital. In practice, there was indeed a steady fall in the rental price of capital (both the interest rates and the relative price of capital fell) in Korea, Taiwan and Hong Kong. The drop is particularly large in Korea, where the national account statistics also suggest a large increase in the capital stock. However, in Singapore, there is no evidence that the rental rate declined over the period. If any thing, it seems to have increased. As for investment in physical capital, there is no prima facie evidence that poor countries invest more in education. The data is poor and extremely partial, since it is difficult to estimate private expenditure on education. What we can measure easily, government expenditure on education as a fraction of GDP, however, is not higher in poor countries, though there is significant variation across countries. In 1996, according to the country level data disseminated by the World Bank “edstat” department, government investment on education was 4.8% in Africa, 4% in Asia, 4.1% in Latin America, 4.8% in North America and 5.6% in Europe. The correlation between the log of government expenditure on education as a fraction of GDP and GDP-per-capita is strong (in current prices): The coefficient of the log of GDP was 0.18 in 1990, and 0.08 in 1996, larger than the comparable estimate for rate of investment in physical capital. As we noted earlier, the fact that teachers are relatively more expensive in developing countries may imply that true returns to education may be much lower than the Mincerian returns. Can this explain why there is not greater investment in education in poor countries? Within the neo-classical model, the answer is no: Banerjee (2004) shows that in the neo-classical world the same forces that raise the relative price of teachers in poor countries (or in countries with low education levels) also raise the wages paid to educated people, and on net the rate of return has to be higher rather than lower. And, in any case, it is not true that public investment in education is higher when returns are higher: We found no correlation between government expenditure on education as a fraction of GDP and rate of returns to education (the coefficient of the rates of return to education on government expenditure in education in 1996 is −0.008, with a standard error of 0.013). In summary, while there are isolated cases of high investment rates in relatively poor countries (Taiwan and Korea), this by no means seems to be a general phenomenon. We have already suggested one reason why this might be the case – it does not look like returns are especially high. It may also be that investment is not particularly responsive with respect to returns. This is the issue we turn to next.
Ch. 7: Growth Theory through the Lens of Development Economics
495
2.2.2. Does investment respond to rates of return? There is little doubt that people do take up many investment opportunities with high potential returns. Investment flowed into Bangalore when it became a hub for the software industry in India. When, in the 1990s, Tirupur, a smallish town in South India, became known in the U.S. as a good place to contract large orders of knitted garments, the industry in the city grew at more than 50% per year, due to substantial investments of both the local community (diversifying out of agriculture) and outsiders attracted to Tirupur [Banerjee and Munshi (2004)]. Or, to take a last example from India, new hybrid seeds and fertilizers spread rapidly during the “green revolution”, leading to very rapid yield growth (yields were multiplied by 3 in Karnataka and 2.5 in Punjab [Foster and Rosenzweig (1996)]). However, there are many instances where investments options with very high rates of returns do not seem to be taken advantage of. For example, Goldstein and Udry (1999) find that, despite the high rates of returns to growing pineapple compared to other crops, only 18% of the land is used for pineapple farming. Similarly, Duflo et al. (2003) find that only less than 15% of maize farmers in the area where they conducted field trials on the profitability of fertilizer report having used fertilizer in the previous season, despite estimated rates of return in excess of 100%. From a more macro perspective, Bils and Klenow (2000) argue that the observed high correlation between educational attainment and subsequent growth observed in cross-sectional data (one year of additional schooling attainment is associated with 0.30 percent faster annual growth over the period 1960–1990) must be due, at least in part, to the fact that higher expected growth rates increase the returns to schooling, and therefore the demand for schooling. As we noted earlier, the correlation between education and subsequent growth [found in many studies, e.g., Barro (1991), Benhabib and Spiegel (1994), and Barro and Sala-i-Martin (1995)] appears to be too high to be entirely explained by the causal effect of transitional differences in human capital growth rates on growth rates. Bils and Klenow (2000) calibrate a simple neo-classical growth model, which requires that the impact of schooling on individual productivity has to be consistent with the average coefficient obtained from Mincerian regressions. Their calibration suggest that the high level of education in 1960 can only explain up to a third of the correlation between education and growth. Moreover, as we will discuss below, this correlation cannot be explained by high human capital externalities. They therefore calibrate an alternative model, where they construct the optimal schooling predicted by a country’s expected economic growth. The calibration, once again, requires that the impact of education on human capital be consistent with the micro-estimates of the Mincerian returns, so that there remains a large fraction of the correlation between education and growth to explain. Higher expected growth induces more schooling by lowering the effective discount rate. They assume that a country’s expected growth is a weighted average of its real ex post growth and the growth of the rest of the world. They estimate that, starting at 6.2 years of schooling, a 1 percent increase in growth induces 1.4 to 2.5 more years of schooling, depending on the values chosen for the parameters
496
A.V. Banerjee and E. Duflo
that are imposed. A 1 percentage point higher Mincerian return to schooling increases education by 1.1 to 1.9 years. The aggregate data is thus consistent with a strong response of schooling to growth. However, it is also consistent with the presence of an omitted variable explaining both education and growth: In fact, Bils and Klenow acknowledge that their estimates suggest an elasticity of schooling demand to returns to schooling that is higher than what is implied by existing micro-studies [reviewed by Freeman (1986)]. This problem cannot really be adequately addressed in the macroeconomic data, since there it is difficult to find a plausible instrument for growth, and the impact of expected growth on schooling must essentially be estimated as a residual impact (what remains to be explained from the correlation between growth and schooling after a plausible estimate for the impact of education on growth has been removed). Foster and Rosenzweig, in a series of papers, use the green revolution in India as a source of partly exogenous increase in rate of returns to human capital to estimate the impact of expected growth and increases in returns to education on schooling and, more generally, investment in human capital. Foster and Rosenzweig (1996) find that returns to education increased faster in regions where the green revolution induced faster technological change: Their estimates imply that in 1971, before the start of the green revolution, the profits in households where the head had completed primary education were 11% higher than the profits in households were he had not. By 1982, the profits were 46% higher for districts where the growth rate was one standard deviation above average. They then turn to estimating whether educational choices were also sensitive to the higher yield growth. After instrumenting for yield growth, they find that the impact of technological change on education is indeed substantial: In areas with recent growth in yields of one standard deviation above the mean, the enrollment rates of children from farm households are an additional 16 percentage points (53%) higher, compared to average-growth areas. Foster and Rosenzweig (2000) find that technological growth also affected the provision of schools, benefiting landless households. However, on balance, technological growth seems to lead to lower educational investment by landless households, perhaps because returns to education increase less for them (since they are engaged in more menial tasks) and because the fact that the withdrawal of children of landed households from the labor market increases children’s wages, and thus the opportunity cost of school attendance. Foster and Rosenzweig (1999) consider another measure of investment in children’s human capital, namely child survival. They argue that technological growth in the village increases the returns from investing in boys’ health, while technological growth outside the village, but in the potential “marriage market”, increases the returns to investing in girls (because better educated and healthier women will fetch a higher prices in regions with higher technological progress). Their results indeed suggest that the gap in boys/girls mortality rates increases with technological change in the village, but decreases with technological change in the labor market. Other evidence that girls’ survival is affected by the expected returns to having girls include Rosenzweig and Schultz (1982), who show that the boys/girls mortality gap is
Ch. 7: Growth Theory through the Lens of Development Economics
497
negatively correlated to women’s wages, and Qian (2003), who uses the liberalization of tea prices in China as a natural experiment in female productivity. She shows that, in regions suitable to tea production, the ratio of boys to girls diminished considerably after tea production and tea prices were liberalized. Since tea is picked by women, this is evidence that higher female productivity encourage parents to invest more in their girls. In contrast, in regions suitable for orchard production (for which males have an advantage) the ratio of boys to girls increased during the period. While these facts taken together do suggest that individuals respond to returns when making human capital investment decisions, there are possible alternative explanations for these facts. The results from Rosenzweig and Schultz (1982) and Qian (2003) cannot easily be distinguished from a women’s bargaining power effect: If mothers tend to prefer girls, and their bargaining power increases as a result of the increase of their productivity, then the outcomes will improve for girls, even if households’ decisions do not respond to returns. The results in Foster and Rosenzweig (1996, 2000) could in part be attributed to wealth effects (expected growth makes the households richer, and if education has any consumption value, one would expect growth to respond to it), although Foster and Rosenzweig (1996) estimate the wealth effect directly, and argue that it is not important. But it remains possible that the instrumented expected increase in yield captures real increases in expected wealth better than any other measure (they show that land prices do adjust to the future expected yield increases, for example). Moreover, there is also direct evidence that investment in human capital does not always respond to returns: Munshi and Rosenzweig (2004) show that the rapid increase in the returns to English education in India in the 1990s (the returns increased from 15% to 24% in 10 years for boys, and 0% to 27% for girls) led to a convergence in the choice of English as a medium of instruction between the low and high castes amongst girls, but not amongst boys: Boys from the lower castes seem so far not to have taken full advantage of the new opportunities offered by English medium education. Another angle for approaching this question is the sensitivity of human capital investment to the direct or indirect costs of these investments. Several recent studies suggest that the elasticity of school participation with respect to user fees is high: Kremer, Moulin and Namunyu (2003) conducted a randomized trial in rural Kenya in which an NGO provided uniforms, textbooks, and classroom construction to seven schools randomly selected from a pool of 14 schools. Dropouts fell considerably in the schools that received the program, relative to the other schools (after five years, pupils initially enrolled in the treatment schools had completed 15% more schooling than those enrolled in the comparison schools). They argue that the financial benefits of the free uniforms were the main reason for this increase in participation. Several programs go beyond reducing the school fees to actually pay for attendance. The PROGRESA program in Mexico provided grants to poor families, conditional on continued school participation and participation in health care. The program was initially launched as a randomized experiment, with 506 communities randomly assigned to either the treatment or control group. Schultz (2004) finds a 3.4% increase in enrollment in all children. The largest in-
498
A.V. Banerjee and E. Duflo
crease was in the transition between primary and secondary school, especially for girls. Gertler and Boyce (2002) report a similar effect on health. In this case as well, it is difficult to distinguish the pure price effect from the income effect.11 School meals, which is another way to pay children to attend school, have been shown to be associated with increased school participation in several observational studies [Jacoby (2002), Long (1991), Powell, Grantham-McGregor and Elston (1983), Powell et al. (1998) and Dreze and Kingdon (2001)] and one experimental trial conducted among pre-school children in Kenya [Vermeersch (2002)]. The available evidence, therefore, points toward a robust elasticity of schooling decisions with respect to the cost of schooling. While this could be indicative of households being extremely sensitive to net returns, the magnitude of these effects is hard to reconcile with this explanation. For example, using an estimate of 7% Mincerian returns per year of education, Miguel and Kremer (2004) estimate that the benefit of one year of primary schooling is in excess of $200 over the lifetime of a child. Yet, the provision of a uniform valued at $6 induced an average increase of 0.5 years in the time a child spent in school (time spent in schools increased from 4.8 years in the comparison schools on average, to 5.3 years in the treatment schools). To be consistent with a model where the only reason where the provision of uniforms increases school attendance is the increase in the rate of returns that it leads to, these numbers would mean that a large fraction of children (or their parents) were exactly indifferent between attending school or not, before the uniform is provided. While this is certainly possible, other evidence suggests that human capital investment does not always respond to rates of returns. For example, the take-up of the de-worming program studied by Miguel and Kremer (2004) was only 57%, despite the fact that the program was free, and that the only investment required was to sign an informed consent form (and some disutility for the child). Further, when a nominal fee was introduced in a randomly selected set of schools in the year after the initial experiment, the take-up fell by 80%, relative to free treatment [Kremer and Miguel (2003)]. While this could be due to the fact that the private benefits are perceived to be low by the parents, it is worth noting that the hike in user fees happened after one year of free treatment, so that parents would have had time to observe the change in the child’s health and attendance at school. Moreover, Kremer and Miguel (2003) also observe that, as long as the price was positive, there was no impact from the actual price on the take-up of the drug. This strong non-linearity between a price of zero and any positive price (which is also consistent with the evidence from school uniforms) appears to be inconsistent with an explanation of their findings in terms of rates of returns. To sum up, the evidence suggests that, while investment seems to respond in part to the cost and the benefits of these investments, it appears to do so in ways that suggest that it does not only respond to returns as we are measuring them.
11 Moreover, there could be a bargaining power effect, since the grants were distributed through women.
Ch. 7: Growth Theory through the Lens of Development Economics
499
2.2.3. Taking stock: investment rates Investment rates, both in physical and human capital, are typically no higher in poorer countries than in rich countries. If we are willing to accept that the average marginal product is the one that guides investment, this is perhaps not a surprise, especially given that the link between investment and rates of return is also not particularly strong.
3. Understanding rates of return and investment rates in poor countries: aggregative approaches For Lucas, the inability to fit the cross-country differences into an aggregative growth model was direct evidence in favor of abandoning the assumption of equal TFP: Allowing the TFP level to be lower in India than in the U.S. will increase the difference in output per worker between the two countries, for any given difference in factor endowments. This is also the message of the more recent literature on development accounting [Klenow and Rodriguez-Clare (1997), Caselli (2005)], which demonstrates that it is impossible to explain even half of the cross-country variation in output per worker based on variation in the stocks of capital and human capital, even after making adjustments for the quality of these inputs, and other possible sources of mismeasurement. Allowing TFP differences across countries can also explain why the poor countries do not invest more and ultimately why there is no growth convergence: Once we assume fixed productivity differences, the steady states in different countries will be different and there is no presumption that poorer countries should grow faster. For rest of this section, we will therefore proceed under the assumption that it is possible to resolve the “macro puzzles” by introducing cross-country differences in TFP. We focus on the so-called new growth approaches, which are theories within the aggregative growth framework that aim to explain persistent cross-country TFP differences, recognizing that these “new” growth theories, like “old” growth theories, make no attempt to deal with the obvious problems with the aggregate production function. 3.1. Access to technology and the productivity gap The dominant answer, within growth theory, of why TFP should be lower in poorer countries comes down to technology. There is a now a large literature – due to Aghion and Howitt (1992), Grossman and Helpman (1991) and others – that emphasizes technological differences as the source of this TFP gap. It is easy to think of reasons why there may be a persistent technology gap between rich and poor countries. Essentially, it is too costly for the poor country to jump to the technological frontier because the frontier technologies belong to firms in the rich countries (who are the ones who have the biggest stake in keeping the technological frontier moving) and they charge monopolis-
500
A.V. Banerjee and E. Duflo
tic prices for access to these technologies.12 Moreover, there is the issue of appropriate technology: The latest technology may not be suitable for use in a country with little human capital13 or poor infrastructure. By itself, this explanation focuses on investment in technology and cannot directly account for the lack of investment in human capital in LDCs or why the returns there often seem so low. However, if there is strong complementarity between human capital investment and investments in new technology,14 then the slow growth of TFP could explain the relative absence of investment in human capital in LDCs, assuming that we accept the rather mixed evidence, reviewed above, on the responsiveness of investment in human capital to the expected returns. If the productivity gap between the U.S. and India has to be fully accounted for by technological differences in an aggregative model (i.e., if we rule out any differences in the interest rates), then TFP in the U.S. would have to be about twice that of India. How plausible is a TFP gap of 1 : 2 in a world of efficiently functioning markets? One way to look at this is to observe that U.S. TFP growth rates seem to be on the order of 1–1.5% a year. Even at 1.5%, TFP takes about 45 years to go up by 200%.15 Therefore in 2000, Indians would have been using machines discarded by the U.S. in the 1950s. This is also clearly very far from being true of the better Indian firms in most sectors. The McKinsey Global Institute’s [McKinsey Global Institute (2001)] recent report on India, reports on a set of studies of the main sources of inefficiency in a range of industries in India in 1999, including apparel, dairy processing, automotive assembly, wheat milling, banking, steel, retail, etc. In a number of these cases (dairy processing, steel, software) they explicitly say that the better firms were using more or less the global best practice technologies wherever they were economically viable. The latest (or if not the latest, the relatively recent) technologies were thus both available in India and profitable (at least for some firms). However, most firms do not make use of these technologies. And, according to the same McKinsey report, it is not because these technologies are not economically viable in this sector: The report on the apparel industry tells us that in the apparel industry: “Although machines such as the spreading machine provide major benefits to the production process and are viable even at current labor costs, they are extremely rare in domestic (i.e., non-exporting) factories” [McKinsey Global Institute (2001)]. 12 The dominance of rich countries in the latest technologies is reinforced by the fact that the rich countries
may have an actual advantage in R&D, because of their larger market size or their superior human capital endowments. 13 For example, as in Acemoglu and Zilibotti (2001) or Howitt and Mayer (2002). 14 As suggested, for example, in the work by Foster and Rosenzweig (1995) on the green revolution, which we discussed above. 15 The effective rate of technological improvement will be larger, for example, if new technology needs to be embodied in machines and machines are more expensive (or savings rates are lower) in the poorer country [see Jovanovic and Rob (1997)].
Ch. 7: Growth Theory through the Lens of Development Economics
501
Despite this, technological backwardness is not one of the main sources of inefficiency that is highlighted in their report on the apparel industry. They focus, instead, on the fact that the scale of production is frequently too small, and in particular, on the fact that the median producer is a tailor who makes made-to-measure clothes at a very small scale, rather than a firm that mass produces clothes. TFP is low, not because the tailors are using the wrong technology given their size, but because tailoring firms are too small to benefit from the best technologies and therefore should not exist. Reports from a number of other industries show a similar pattern. Certain specific types of technological backwardness are mentioned as a source of inefficiency in both the dairy processing industry and the telecommunications industry, but in both cases it is argued that while all firms should find it profitable to upgrade along these dimensions [McKinsey Global Institute (2001)], only a few of them do. In these two cases, however, there is also a reference to the gains (in terms of productive efficiency) from what the report calls “non-viable automation”. This is automation that would raise labor productivity but lower profits. One reason why automation may be non-viable in this sense is that the technology may be under patent and therefore expensive, along the lines suggested by Aghion and Howitt (1992), Grossman and Helpman (1991) and others, or it may demand skills that the country does not have. On the other hand, it could also be something entirely neo-classical: Labor-saving devices are less useful in labor-abundant countries. Since we have no way of determining why the technology is non-viable, we looked at the total labor productivity gain promised by this category of innovations. In both the dairy processing industry and the telecom case, this number is 15% or less, and in the automotive industry it is no larger [McKinsey Global Institute (2001)]. This is clearly nowhere near being large enough to explain the entire TFP gap. On other hand, it is clearly true that there are many firms that, for some reason, have opted not to adopt the best practice despite the fact that others within the same economy find it profitable to do so and, at least according to McKinsey, they too would benefit from moving in this direction. In other words, while there is a technology gap, it is largely a within-country phenomenon and not, as the models of technology production and adoption imply, a problem at the level of the country.16 3.2. Human capital externalities Another reason why there may be persistent TFP differences across countries is that there are aggregate increasing returns. As emphasized in the introduction, for this to be true it is not enough to have firm-level increasing returns. We need externalities across firms, or more generally across investors, which may arise, for example, because there 16 It is true that machines of different vintages, and therefore different productivities, can co-exist even when
markets work perfectly, as long as machines are long-lived [Bardhan (1969)]. However the technological gap between the tailor and the garment factory (or the milk-man and the fully-automatized dairy) seem too large and too stable to be just a transitional problem.
502
A.V. Banerjee and E. Duflo
are human capital externalities: It has been argued that human capital is not just valuable to those who own and use it, but also to others. Externalities in human capital would tend to limit the extent of diminishing returns with respect to human capital in the production function, keeping as given the share of human capital in total production. This would tend to raise productivity in rich countries (who have a lot of human capital) and slow down convergence. Externalities could also explain a puzzle we did not discuss until now, pointed out by Acemoglu and Angrist (2001) and Bils and Klenow (2000): The high correlation between human capital and income that is observed in the cross-country data [e.g., Mankiw et al. (1992)] is hard to reconcile with the micro evidence we have reviewed earlier, which suggested relatively low returns to education. To see this, note that the difference in average schooling between the top and bottom deciles of the world education distribution in 1985 was less than 8 years. With a Mincerian returns to schooling of about 10%, the top decile countries should thus produce about twice as much per worker as the countries in the bottom decile. In fact, the output-per-worker gap is about 15. One possibility is that the Mincerian rate of return understates the true rate of returns to education, because it does not take into account positive externalities generated by educated workers. More specifically, the human capital externalities on the order of 20–25% (more than twice the private return) would be necessary to explain the crosscountry relationship between education and income, which sounds implausible. Early evidence [e.g., Rauch (1993)] suggested that externalities were positive, but not of that order of magnitude. Using variation in education across U.S. cities, Rauch (1993) estimated that the human capital externalities may be on the order of 3% to 5%. Moreover, even this evidence is to be taken with caution, since cities where workers are more educated vary in many other respects. Using variation in average education generated by the passage of compulsory schooling laws, Acemoglu and Angrist (2001) find no evidence of average education on individual wages, after controlling for individual education. In Indonesia, Duflo (2003) actually finds evidence that those who invest in their education may inflict negative pecuniary externalities on others. She studies the impact of an education policy change that differentially affected different cohorts and different regions of Indonesia. Between 1973 and 1979, oil proceeds were used to construct over 61,000 primary schools throughout the country. Duflo (2004) shows that the program resulted in an increase of 0.3 years of education for the cohorts exposed to the program. Duflo (2003) takes advantage of the fact that individuals that were 12 or older when the program started did not benefit from the program, but worked in the same labor markets as those who did. As the newly educated workers entered the labor force, starting in the 1980s, the fraction of educated workers in the labor force increased. Since migration flows in Indonesia remained relatively modest, the increase in the fraction of workers with primary education between 1986 and 1999 was faster in regions which received more INPRES schools. Using the interaction of year and region as instruments for the fraction of educated workers, she estimates that an increase of 10 percentage points in the fraction of educated workers in the labor force resulted in a decrease in the wages
Ch. 7: Growth Theory through the Lens of Development Economics
503
of the older workers (both educated and uneducated) by 4% to 10%. This suggests that, on balance, there are strongly diminishing aggregate returns at the local level: Any positive externality is more than compensated by these declining returns. The Mincerian returns could then actually overestimate the aggregate returns of increasing education, because by comparing individuals within a labor market, they do not take into account the diminishing returns that affect everybody in the labor market. To summarize, the available evidence does not suggest that there are strongly increasing returns to human capital. It appears that if human capital externalities are important they must take a very different form. One possibility is that they play a role in the intergenerational transmission of learning: For example, it is possible that parents or teachers do not fully internalize the benefits that their investment in human capital confer on the next generation of scholars. However as the figures above makes clear, in order to fit the data the extent of this miscalculation has to be substantial. 3.3. Coordination failure Another source of lower aggregate productivity is the possibility of coordination failures, which reduces aggregate productivity through a demand effect. There is a long line of work, starting with Rosenstein-Rodan (1943), that has emphasized the role of coordination failure in explaining why certain countries successfully industrialize, while others remain poor and non-industrialized. Murphy, Shleifer and Vishny (1989) explore models where industrialization in a sector creates demand for the products of another sector (through higher wages for the workers), and which leads to multiple equilibria. A coordinated “big push”, where all industries start together, can place the country on a permanently higher level of investment and income. Developing countries may have low investments and low returns to capital because such a “big push” has not happened. A large literature explores different forms of strategic complementarities. Since the argument involves an entire economy’s coordination, it is difficult to use micro-evidence to provide much direct evidence about these aggregate externalities.17 However, while these theories certainly have some relevance, the fact countries trade will tend to substantially mitigate the effect of local demand. It is therefore not clear that aggregate demand effects can be so powerful as to generate the necessary gap in TFP between, say, India and the U.S. Another possible mechanism is suggested by Acemoglu and Zilibotti (1997). They argue that when one firm invests, others benefit, because each firm is subject to independent shocks and increasing the number of firms expands the ability of an individual firm to diversify its risk. When there are few firms around, risk diversification opportunities are limited and risk averse investors limit their investment. The India–U.S. comparison is perhaps the worst example one could pick for applying this theory, since between the 1920s and the 1940s the Indian stock market was
17 Below, we will review the evidence on more local externalities.
504
A.V. Banerjee and E. Duflo
comparable to that in many OECD countries in terms of number of listed firms and volume of trade. However, lack of financial development is clearly a serious problem for many developing countries. However, within an aggregative model the lack of financial development can at best explain why all the firms underinvest. As emphasized earlier, the more important sources of inefficiency seem to come from the fact that some firms underinvest much more than others, and in particular that some firms adopt the latest technologies, while others do not. 3.4. Taking stock While the evidence is somewhat impressionistic, it seems unlikely that the aggregative theories discussed above can explain the entire TFP gap. Of course, if we were prepared to give up the idea that the entire problem comes from a lower aggregate productivity, for example by accepting that the marginal product is lower in India, the problem of fitting the data would be easier. For example, if the TFP gap were 1.5 higher in the U.S. (on top of what is predicted by the difference in the productivity of labor), the fact that the U.S. has 18 times more capital-per-worker would imply that output-per-worker would be (1.5)(2)(18)0.4 = 9.5 times higher in the U.S., and the marginal product 0.6 of capital would be (18) 2(1.5) = 1.9 times higher in India. These are both clearly in the ballpark, although the output gap between the U.S. and India predicted by this model is still too low (the output gap is about 11 : 1 in the data) and the ratio of the marginal product of capital between India and the U.S., which was too high in a model with identical TFP, is now too low (the ratio in the data is about 2.5). It is worth noting that in order to get closer to 11 : 1 ratio in the data, the TFP ratio would need to be higher than 1.5, which is perhaps already too big. Moreover, this would further reduce the predicted ratio between the marginal product of capital in India and in the U.S., which was already too low when the TFP gap was 1.5. In other words, we are facing a new problem: Given the existing capital stock, if a difference in TFP was the reason why the output per worker is so low in India, the marginal product of capital should be even lower in India than what it is. Indeed, there is no way to adjust the TFP ratio to improve the fit along both dimensions – we can increase the gap in output-perworker by raising the TFP ratio, but only at the cost of making the ratio of marginal product even smaller. The problem is quite basic: With a Cobb–Douglas production function, the average product of capital is proportional to its marginal product. But then output-per-worker must be proportional to the product of the marginal product of capital and capital-per-worker. If the marginal product in India is 2.5 times that of the U.S., but 18 capital-per-worker is 18 times greater in the U.S., output-per-worker has to be 2.5 = 7.2 times larger in the U.S. and not 11 times larger, irrespective of what we assume about the ratio of TFP in the two countries. In other words, the only way we can hope to really fit what we see in the data is by abandoning the standard Cobb–Douglas formulation. This is useful to keep in mind when, in later sections, we discuss ways to improve the fit between the theory and the data.
Ch. 7: Growth Theory through the Lens of Development Economics
505
To sum up, Lucas’ question about why capital does not flow from the U.S. to India was, in some sense, where it all started, but from the vantage point of what we know today, this is in some ways the lesser problem. We know now that there are differences in the marginal product of capital within the same economy that dwarf the gap that Lucas calculated from comparison of India and the U.S., and found so implausibly large that he set out to rewrite all of growth theory. The harder question is why capital flows do not eliminate these differences. Lucas’ resolution of the puzzle was to give up the key neo-classical postulate of equal TFP across countries. Based on the McKinsey report, this seems to be the obvious step, but the problem is less that people in developing countries do not find it profitable to adopt the latest (and best) technologies and more that many firms do not adopt technologies that are available and would be profitable if adopted. The key question, once again, is why the market allows this to be the case. The premise of the aggregative approach to growth was that markets function well enough within countries that we can largely ignore the fact that there is inefficiency and unequal access to resources within an economy when we are interested in dynamics at the country level. The evidence suggests that this is not true: The cross-country differences in marginal products or technology that we want to explain are of the same order of magnitude as the differences we observe within each economy. A theory of cross-country differences has to based on an understanding and an acknowledgment of the reasons why rates of returns vary so much within each country. This is what we turn to next: In Section 4, we first review the various reasons that have been proposed. In Section 5, we will then calibrate their impact to evaluate whether they can form the basis of an explanation for the puzzles we observed.
4. Understanding rates of return and investment rates in poor countries: non-aggregative approaches In this section, we review various possible reasons why individuals do not always make the best possible use of resources available to them. 4.1. Government failure One reason why firms may not choose the latest technologies or make the right investments is because they do not have the proper incentives to do so. A line of work has developed the hypothesis that governments are largely responsible for this situation, either by not protecting investors well enough or by protecting some of them excessively. The firms that are ill-protected underinvest and have high marginal returns, while the over-protected firms overinvest and show low marginal returns. The net effect on investment may be negative, because even those who are currently favored may fear a future falling out and a corresponding loss of protection. Overall productivity may also
506
A.V. Banerjee and E. Duflo
go down, since the right people may not always end up in the right business, since connections rather than skills will dominate the choice of professions.18 One approach to investigating this hypothesis has been to try to document variations in the quality of institutions, and to try to evaluate their impact. La Porta et al. (1998) document important variations in the degree to which the law protects investors (creditors and shareholders) across countries, part of which seem to be explained by the origin of these countries’ legal codes (the French civil law has much less legal protection for investors than Anglo-Saxon common law). Djankov et al. (2002) document wide variation in the ability of someone to start a new firm in 85 countries. They argue that the costs of entry are high in most countries (on average, they sum up to 47% of a country’s GDP per capita), and can be very high indeed: While it take 3 procedures and 3 days to obtain the permit to start company in New Zealand, it takes 19 procedures, 149 business days and 111.5 percent of GDP per capita in Mozambique. The procedure is shorter, and generally less expensive in terms of GDP per capita, in rich countries than in poor or middle-income countries. Djankov et al. (2003) document the time it takes in court to evict a tenant or collect a bounced check, as well as the degree of formalism of the legal procedures. They find, once again, wide variation: In particular, these procedures take a much shorter time in countries with common law legal origins. Similarly, many studies argue that, in cross-country regressions, there is a strong association between aggregate investment and measures of bad institutions or corruption [e.g., Knack and Keefer (1995), Mauro (1995), Svensson (1998)]. These papers also argue that low investor protections, legal barriers to entry, and long legal procedures have implications for welfare and efficiency. There are indeed suggestive associations in the data (for example, ownership is more concentrated when investor protection is worst), but there is always the possibility that the correlation between the quality of the institutions and the real outcomes they consider is due to a third factor. Acemoglu, Johnson and Robinson (2001) try to address this issue by finding exogenous variations in the quality of institutions. They argue that there is a persistence of institutions, so that countries which accessed independence with extractive institutions (e.g., Congo) have tended to keep these bad institutions. They then argue that colonial powers were more likely to set up extractive institutions, with an unrestrained executive power, in places where they did not intend to settle. Finally, they were less likely to settle in places where the environment was hostile: In particular, the mortality of early settlers predicted the number of people of European descent who settled in these countries, the quality of institutions at the turn of the 20th century, and the quality of institutions in 1995 (measured as the risk of expropriation perceived by investors). In turn, it also is associated with lower GDP in 1995. The authors then use early settler mortality as an instrument for institutions in a regression of the impact of institutions on inequality, and find a strong positive coefficient. This evidence suggests that governments matter, and that bad governments will lower returns and discourage new investments. There is a literature that tries to investigate the 18 See Murphy, Shleifer and Vishny (1995).
Ch. 7: Growth Theory through the Lens of Development Economics
507
exact mechanisms through which the government affects the allocation of resources. One version of the story blames excessive intervention, while another talks about the lack of appropriate regulations. We now discuss these two explanations in turn, and try to assess how far they can help us fit the evidence. 4.1.1. Excessive intervention There is a line of work, following Parente and Prescott (1994, 2000), which argues that the productivity gap results from the way the heavy hand of the government operates. The government makes rules that discourage entry and innovation and protects the inept, and thereby slows the economy’s progress towards the ideal state where only the most productive firms survive. The regulation may lead the economy to have too few firms, leaving inefficient incumbents to run the firms [see Caselli and Gennaioli (2004) and other references in this study], or too many firms, when regulation discourages consolidation by treating small firms and larger firms differently. There is clearly something to this vision. Gelos and Werner (2002) show that financial de-regulation in Mexico (which started in 1988 and eliminated the interest rate ceiling, high reserve requirements which channeled 72% of commercial bank lending to the government, and priority lending) increased the ability of small firms to access the credit market, and reduced the excess cash flow sensitivity of investment for small firms only. Until recently in India, a large number of sectors were reserved for firms below a certain size (the small-scale sector) and/or firms in the cooperative sector. Small firms also benefited from tax exemptions and priority sector credits. This clearly limited the ability to take advantage of economies of scale and restricted the market share of the most efficient players. Nonetheless, this is probably only a part of the story. As we noted in the context of the discussion of Banerjee and Duflo (2004), even medium-sized firms that were well above the cut-off for being included in the small-scale sector seem to be operating well below their optimal scale. In other words, notwithstanding the politically protected presence of the small-scale firms that is presumably driving down profits in the sector, these medium-sized firms were clearly still at the point where further investment would be extremely profitable. There has to be something other than a policy-induced lack of profitability that was holding them back. The same point is made in a different style in the paper by Banerjee and Munshi (2004), mentioned above. This paper studies investment and productivity differences among firms in the knitted-garment industry in Tirupur, India. The firms owned by the Gounders tend to be much larger than the firms owned by all other participants in the industry: The gap among firms that had just started is on the order of three to one. Yet these Gounder firms produce much less per unit of capital, and Gounder firms that have been in business for more than five years actually produce less in absolute terms than the smaller firms of the same vintage owned by non-Gounders. In other words, it is the bigger firms that are less productive, in an environment where the government discriminates, if at all, in favor of the smaller firms.
508
A.V. Banerjee and E. Duflo
To sum up, while there are certainly instances of excessive intervention, it seems that there are many inefficiencies that cannot be blamed on the government. 4.1.2. Lack of appropriate regulations: property rights and legal enforcement Effective rates of return and investment rates can be low because the responsibilities and/or the benefits of the investments are shared, or the investors are worried about being expropriated: The investor is therefore not capturing the full marginal returns of its investment. Imperfect property rights will thus lead to low investments. Poorly enforced property rights also make it difficult to provide collateral, which exacerbates the problems of the credit market. For example, the study of the Mexican financial deregulation discussed above [Gelos and Werner (2002)] showed that after the deregulation, small firms’ access to credit became more linked to the value of the real estate assets they could use as collateral: The role of the government does not end with not interfering, it may also be to provide secure property rights. In addition to the macro-economic evidence mentioned above, there is some microeconomic evidence that property rights matter for investment, although the findings are more mixed. Goldstein and Udry (2002) show that, in Ghana, individuals are less likely to leave their land fallow (which is an investment in long run land productivity) if they do not hold a position of power within the family of the village hierarchy which ensures that their land is not taken away from them when it is fallow. However, Besley (1995) finds that, also in Ghana, investment (tree planting) is not significantly larger when individuals have more secure rights to their land. Johnson, McMillan and Woodruff (2002) find that, in five post-Soviet countries, firms that are run by entrepreneurs who perceive that their property rights are more secure invest more than those who do not. The effect is as strong for firms who rely mostly on internal finances as for those who have access to external finance. Entrepreneurs who believe that they have strong property rights invest 56% of their profits in their firms (against 32% for those who do not). Do and Iyer (2003) find that a land reform which gave farmers the right to sell, transfer or inherit their land usage rights also increased agricultural investment, in particular the planting of multi-year crops (such as coffee). Even when property rights themselves are legally well defined and protected, there are institutions which reduce the private incentives to invest. Sharecropping is one environment where both the landlord and the tenants have low incentive to invest in the inputs that they are responsible for providing [Eswaran and Kotwal (1985)]. Binswanger and Rosenzweig (1986) and Shaban (1987) both show that, controlling for farmer’s fixed effect (that is, comparing the productivity of owner-cultivated and farmed land for farmers who cultivate both their own land and that of others) and for land characteristics, productivity is 30% lower in sharecropped plots. Shaban (1987) shows that all the inputs are lower on sharecropped land, including short-term investments (fertilizer and seeds). He also finds systematic differences in land quality (owner-cultivated has a higher price per hectare), which could in part reflect long-term investment. Banerjee, Gertler and Ghatak (2002) study a tenancy reform which increased the tenants’ bargaining power
Ch. 7: Growth Theory through the Lens of Development Economics
509
and security of tenure. They found that the land reform resulted in a substantial increase in the productivity of the land (62%). Since the reform took place at the same time as the green revolution, this increase in productivity is probably in part due to an increased willingness to switch to the new seeds after the registration program.19 The example of sharecropping suggests that bad governments are not the only cause for the emergence of bad institutions. If sharecropping is inefficient, why does it arise? In particular, why do the landlord and the tenant not agree on a fixed rent, which will ensure that the tenant is the full beneficiary of his effort at the margin? Explanations of the persistence of sharecropping involve risk aversion [Stiglitz (1974)] or limited liability [Banerjee et al. (2002)]. This suggests that while the proximate explanation for inefficient investment may well be based in a specific institution, the more basic cause may be lying elsewhere, in the way various asset markets function. This is what we turn to next. 4.2. The role of credit constraints • Why would credit markets function poorly in poor countries? The fact that the capital market does not function well in poor countries is a result of a number of factors. First, information systems, including property records, are often underdeveloped, making it hard to enforce contracts. This, in turn, partly reflects the fact that people may not know how to read or write and partly the fact that there has not been enough institutional investment.20 Second, the fact that potential borrowers are poor and under extreme economic pressure, might make them all too willing to try to cheat the lender. Third, there are political pressures to protect borrowers from lenders in most LDCs. • Consequence of poorly functioning credit market Given the problems in enforcing the credit contract, what a lender will be prepared to offer a particular borrower will depend on the quality of the borrower’s collateral, his reputation in the market, the ease of keeping an eye on him and a host of other characteristics of the borrower. This has the obvious implication that two firms facing the exact same technological options may end up choosing very different methods of production. In particular, one person may start a large or more technologically advanced firm because he has money and another may start a small and backward one because he
19 This interpretation is reinforced by the fact that their estimates are higher than Shaban’s and that of a study
by Laffont and Matoussi (1995) who use data from Tunisia to show that a shift from sharecropping to fixedrent tenancy or owner cultivation raised output by 33 percent, and moving from a short-term tenancy contract to a longer-term contract increased output by 27.5 percent. 20 For example, Djankov et al. (2002) document the time it takes the recover a bounced check across countries. It takes longer in poorer countries.
510
A.V. Banerjee and E. Duflo
does not.21 As a result, neither interest rates, nor TFP, nor the marginal product need be equalized across borrowers. This would also explain why investment responds so unpredictably to returns: Sometimes the opportunities become available when there is large group of people who are looking to invest and have the wherewithal to do it. At other times, the returns may be there but most of those who have money may be heavily involved in promoting something else. A second set of implications of imperfect contracting in the credit market is that the supply curve of capital to the individual borrower slopes up – a borrower who is more leveraged will need more monitoring and the lender will charge him more to do the extra monitoring. And eventually, the extra monitoring may be too costly to be worth it, and the borrower will face an absolute limit on how much he can borrow. An immediate consequence of an upward-sloping supply curve is that the marginal product of capital will be higher than what the borrower pays the lender. Indeed, the gap between the two may quite substantial, since the fact that borrowers are constrained in borrowing also implies that the lenders are constrained in how much they can lend at rewarding rates. This drives the interests rates down, as lenders compete for the best borrowers. Moreover, since the rates the lenders charge include the cost of the monitoring that they have to do, the rates the lenders charge could be much higher than the opportunity cost of capital. In the case of a financial intermediary, such as a bank, this implies that the rates they charge their borrowers may be much higher than the rates they pay their depositors. This implies, for example, that the American investor who gets 9% on his stock market investment could not just put the money in a bank in India and earn the 22% average marginal product. Indeed, he may not earn much more than 9% if he were to put it in an Indian bank. However, he could set up a business in India and earn those returns, and presumably if enough people did that, the returns would be equalized; below we will try to say something about why this does not happen. It also implies that the incentive to save may be low in countries where the marginal product is high, except for those who are planning to invest directly. This might help to explain the low equilibrium investment rate, though it is theoretically possible that the negative effect on the savers would be swamped by the positive effect on investors if the fraction of investors is large enough. • Evidence We have already mentioned some evidence from South Asia showing that the interest rate varies enormously across borrowers within the same local capital market and that the extent of variation is too large to be explained by the observed differences in default rates. Banerjee (2003) lists a number of studies that make it clear that this is also true in
21 Aghion, Howitt and Mayer-Foulkes (2004) argue that credit constraints may also be important in explain-
ing the cross-country differences in the adoption of new technologies.
Ch. 7: Growth Theory through the Lens of Development Economics
511
developing countries outside South Asia. This is suggestive, albeit indirect, evidence of credit constraints. If the marginal product of capital in the firm is greater than the market interest rate, credit constraints naturally mean that a firm would want to borrow more than what is available. It is, however, not clear how one should go about estimating the marginal product of capital. The most obvious approach, which relies on using shocks to the market supply curve of capital to estimate the demand curve, is only valid under the assumption that the supply is always equal to demand, i.e., if the firm is never credit constrained. The literature has therefore taken a less direct route: The idea is to study the effects of access to what are taken to be close substitutes for credit – current cash flow, parental wealth, community wealth – on investment. If there are no credit constraints, greater access to a substitute for credit would be irrelevant for the investment decision. While this literature has typically found that these credit substitutes do affect investment,22 suggesting that firms are indeed credit constrained, the interpretation of this evidence is not uncontroversial. The problem is that access to these other resources is unlikely to be entirely uncorrelated with other characteristics of the firm (such as productivity) that may influence how much it wants to invest. To take an obvious example, a shock to cash-flow potentially contains information about the firm’s future performance. The estimation of the effects of credit constraints on farmers is significantly more straightforward since variation in the weather provides a powerful source of exogenous short-term variation in cash flow. Rosenzweig and Wolpin (1993) use this strategy to study the effect of credit constraints on investment in bullocks in rural India. The paper by Banerjee and Duflo (2004) that we discussed above makes use of an exogenous policy change that affected the flow of directed credit to an identifiable subset of firms in India. Since the credit was subsidized, an increase in sales and investment as a response to the increase in funds available needs to mean that firms are credit constrained, since it may have decreased the marginal cost of capital faced by the firm. However, they argue that if a firm is not credit constrained, then an increase in the supply of subsidized directed credit to the firm must lead it to substitute directed credit for credit from the market. Second, while investment, and therefore total production, may go up even if the firm is not credit constrained, it will only go up if the firm has already fully substituted market credit with directed credit. They showed that bank lending and firm revenues went up for the newly targeted firms in the years when the priority sector was expanded to include them, and declined in the years where they were excluded again. They find no evidence that this was accompanied by substitution of bank credit for borrowing from the market and no evidence that revenue growth was confined to firms that had fully substituted bank credit for market borrowing. As already argued, 22 The literature on the effects of cash-flow on investment is enormous. Fazzari, Hubbard and Petersen (1988)
provide a useful introduction to this literature. The effects of family wealth on investment have also been extensively studied [see Blanchflower and Oswald (1998) for an interesting example]. There is also a growing literature on the effects of community ties on investment [see, for example, Banerjee and Munshi (2004)].
512
A.V. Banerjee and E. Duflo
the last two observations are inconsistent with the firms being unconstrained in their market borrowing. The logic of credit constraints applies as much or more to human capital investments. Hart and Moore (1994), among others, have used human capital as the archetype of investment that cannot be collateralized, and therefore is hard to borrow against. This is made even more difficult by the fact that children would need to borrow for their education, or parents would need to borrow on their behalf. We return to this evidence below. The high responsiveness to user fees that we reviewed in Section 2, and the evidence that investment in education are sensitive to parental income,23 are both consistent with credit constraints. However, because human capital investments may involve direct utility or disutility (for example, a parent may like to see his child being educated), it is more difficult to come up with evidence that systematically nails the role of credit constraints for human capital investment. Edmonds (2004) is an interesting attempt to try to isolate the effect of credit constraints using household’s response to an anticipated income shock. He studies the effect on child labor and education of a large old age pension program, introduced in South Africa at the end of the Apartheid. Many children live with older family members (often their grandparents). Women become eligible at age 60 and men become eligible at age 65. Since at the time he studies the program, the program was well in place and therefore fully anticipated, he argues that if more children start attending school as soon as their grandfather or grandmother crosses the age threshold and becomes eligible (rather than continuously, as they come closer to eligibility), this must be an indication of credit constraint. Indeed, he finds that child labor declines, and school enrollment increases, discretely when a household member becomes eligible. • Summary Credit constraints seem to be pervasive in developing countries. Of course, we are interested in whether the fact that access to capital varies across people helps us understand the productivity gap. If people invest different amounts because of differential access to capital, our intuitive presumption would be that capital is being misallocated, because there is no reason why richer people are always better at making use of the capital. This misallocation could be a source of difference in productivity. We will return to this question in Section 5. 4.3. Problems in the insurance markets Even if credit markets function well, and there is no limited liability, individuals may be reluctant to invest in any risky activity, for fear of losing their investment, if they are not properly insured against fluctuations in their incomes. Risk aversion leads to inefficient investment, and efficiency would improve with insurance [this idea is explored
23 See Strauss and Thomas (1995) for several studies along these lines.
Ch. 7: Growth Theory through the Lens of Development Economics
513
theoretically in Stiglitz (1969), Kanbur (1979), Kihlstrom and Laffont (1979), Banerjee and Newman (1991), Newman (1995) and Banerjee (2005)]. • Insurance in developing countries A considerable literature has investigated the extent of insurance in rural areas in developing countries [see Bardhan and Udry (1999) for a survey]. Townsend (1994) used the ICRISAT data, a very detailed panel data set covering agricultural households in four villages in rural India to test for perfect insurance. The main idea behind this test is that with perfect insurance at the village level only aggregate (villagelevel) income fluctuation, and not idiosyncratic income fluctuations, should translate into fluctuation in individual consumption. He was unable to reject the hypothesis that the villagers insure each other to a considerable extent: Individual consumption seems to appear to be much less volatile than individual income, and to be uncorrelated with variations in income. This exercise had limits, however [see Ravallion and Chaudhuri (1997) for a comment on the original paper], and subsequent analyses, notably by Townsend himself, have shown the picture to be considerably more nuanced. Deaton (1997) shows that there is no evidence of insurance in Cote d’Ivoire. Townsend (1995) finds the same results across different areas in Thailand. Fafchamps and Lund (2003) find that, in the Philippines, households are much better insured against some shocks than against others. In particular, they seem to be poorly insured against health risk, a finding corroborated by Gertler and Gruber (2002) in Indonesia. Most interestingly, Townsend (1995) describes in detail how insurance arrangements differ across villages. While in one village there is a web of well-functioning risk-sharing institutions, the situations in other villages are different: In one village, the institutions exist but are dysfunctional; in another village, they are non-existent; finally, in a third village, close to the roads, there seems to be no risk-sharing whatsoever, even within family. This last fact is attributed to the proximity to the city, which makes the village a less close-knit community, where enforcement of informal insurance contracts is more difficult. Coate and Ravallion (1993) was the first paper to build a theoretical model of insurance with limited commitment, and to show that, when the only incentive to contribute to the insurance scheme in good times is the fear of being cut away from the insurance in future periods, insurance will be limited. It will also be optimal to make payment contingent on past history, which will lead to a blur between credit and insurance [Ray (1998)]. Udry (1990) presents evidence from Nigeria that is consistent with this model. The villages he studies are characterized by a dense network of loan exchange: Over the course of one year, 75% of the households had made loans, 65% had borrowed money, and 50% had been both borrowers and lenders. Ninety-seven percent of these loans took place between neighbors and relatives. Most importantly, the loans are “state-contingent”: Both the repayment schedule and the amount repaid are affected by the lender’s state and the borrower’s state. This is evidence that credit is to some extent used as an insurance device. The resulting system is a mix of credit and insurance close to what the model of limited commitment would predict. However, and
514
A.V. Banerjee and E. Duflo
still consistent with this prediction, there is not enough of this “security” to fully insure households against income fluctuations: A shock to a particular borrower has a negative impact on the sum of the transfers received by his lender, which means that the lender did not fully diversify risk. Despite this evidence, we do not fully understand the reasons for the lack of insurance among households. It is unlikely that either limited commitment or the more traditional explanations in terms of moral hazard or adverse selection can explain why the level of insurance seems to vary from one village to the next, or why there is no more insurance against rainfall, for example. • Consequences for investment Irrespective of the ultimate reason for the lack of insurance, it may lead households to use productive assets as buffer stocks and consumption smoothing devices, which would be a cause for inefficient investment. Rosenzweig and Wolpin (1993) argue that bullocks (which are an essential productive asset in agriculture) serve this purpose in rural India. Using the ICRISAT data, covering three villages in semi-arid areas in India, they show that bullocks, which constitute a large part of the households’ liquid wealth (50% for the poorest farmers), are bought and sold quite frequently (86% of households had either bought or sold a bullock in the previous year, and a third of the householdyear observations are characterized by a purchase or sale), and that sales tend to take place when profit realizations are low, while purchases take place when profit realizations are high. Since there is very little transaction in land, this suggests that bullocks are used for consumption smoothing. Because everybody needs bullocks around the same time, and bullocks are hard to rent out, Rosenzweig and Wolpin estimate that, in order to maximize production efficiency, each household should own exactly two bullocks at any given point in time. The data suggest that, for poor or mid-size farmers there is considerable underinvestment in bullocks, presumably because of the borrowing constraints and the inability to borrow and accumulate financial assets to smooth consumption: Almost half the households in any given year hold no bullock (most of the others own exactly two).24 Using the estimates derived from a structural model where household use bullocks as a consumption smoothing device in an environment where bullocks cannot be rented and there is no financial asset available to smooth consumption, they simulate a policy in which the farmers are given a certain non-farm income of 500 rupees (which represents 20% of the mean household food consumption) every period. This policy would raise the average bullock holding to 1.56, and considerably reduce its variability, due to two effects: The income is less variable, and by increasing the income, it makes “prudent” farmers (farmers with declining absolute risk aversion) more willing to bear the agricultural risk.
24 The fact that there is under-investment on average, and not only a set of people with too many bullocks
and a set of people with too few, is probably due to the fact that bullocks are a lumpy investment, and owning more than two is very inefficient for production – there is no small adjustment possible at the margin.
Ch. 7: Growth Theory through the Lens of Development Economics
515
Moreover, we observe only insurance against the risks that people have chosen to bear; the inability to smooth consumption against variation in income may lead households to choose technologies that are less efficient, but also less risky. Banerjee and Newman (1991) argue, for example, that the availability of insurance in one location (the village), while its unavailability in another (the city), may lead to inefficient migration decisions, since some individuals with high potential in the city may prefer to stay in the village to remain insured. There is empirical evidence that households’ investment is affected by the lack of ex post insurance. Rosenzweig and Binswanger (1993) estimate profit functions for the ICRISAT villages, and look at how input choices are affected by variability in rainfall. They show that more variable rainfall affects input choices, and in particular, poor farmers make less efficient input choices in a risky environment. Specifically, a one standard deviation increase in the coefficient of variation of rainfall leads to a 35% reduction in the profit of poor farmers, 15% reduction in the profit of median farmers, and no reduction in the profit of rich farmers. Morduch (1993) specifically investigates how the anticipation of credit constraint affects the decision to invest in HYV seeds. Using a methodology inspired by Zeldes (1989), he splits the sample into two groups, one group of landholders who are expected to have the ability to smooth their consumption, and one group that owns little land, whom we expect a priori to be constrained. He finds that the more constrained group uses significantly less HYV seeds. It is worth noting that the estimated impact of lack of insurance on investment is likely to be a serious underestimate. It is not clear how one could evaluate how much the lack of insurance affects investment. While we might observe certain options considered by the investor, there is no obvious way for knowing what other, even more lucrative, choices he chose not to even think about. Another strategy for looking at the effects of underinsurance is to calculate the effect based on the assumption of specific utility function. This, in effect, is what Krussel and Smith (1998) do. They argue that, for reasonable parameter values, the effect on aggregate investment tends to quite small: This is because most people can self-insure quite well against idiosyncratic shocks, and those who cannot, mainly the very poor, do very little of the investing in any case. However as pointed out by a more recent paper by Angeletos and Calvet (2003), the Krusell and Smith result relies heavily on the assumption that one cannot limit exposure to risk by investing less. If investing more exposes you to more risk, even the non-poor will worry about risk, because they are the ones who invest a lot and therefore are exposed to a lot risk. 4.4. Local externalities As we discussed in Section 4, there is a line of work that focuses on coordination failures at the level of the economy. However, Durlauf (1993) shows that externalities do not have to be aggregated for the economy to exhibit multiple equilibria: Local complementarities (where adoption of a particular technology lowers production costs in a few
516
A.V. Banerjee and E. Duflo
“neighboring” sectors) can build up over time to affect aggregate behavior and generate lower aggregate growth. An example of strategic complementarity of this kind arises when agents are learning from each other. Banerjee (1992) shows how, when people try to infer the truth from other people’s actions, this leads them to under-utilize their own information, and leads to “herd behavior”. While this behavior is rational from the point of view of the individual, the resulting equilibrium is inefficient, and can lead to underinvestment, overinvestment, or investment in the wrong technology.25 The impact of learning on technology adoption in agriculture has been studied particularly extensively. Besley and Case (1994) show that in India, adoption of HYV seeds by an individual is correlated with adoption among their neighbors. While this could be due to social learning, it could also be the case that common unobservable variables affect adoption of both neighbors.26 To partially address this problem, Foster and Rosenzweig (1995) focus on profitability. As we mentioned previously, during the early years of the green revolution, returns to HYV were uncertain and dependent on adequate use of fertilizer. In this context, the paper shows that profitability of HYV seeds increased with past experimentation, by either the farmers or others in the village. Farmers do not fully take this externality into account, and there is therefore underinvestment. In this environment, the diffusion of a new technology will be slow if one neighbors’ outcomes are not informative about an individual’s own conditions.27 Indeed, Munshi (2004) shows that in India, HYV rice, which is characterized by much more varied conditions, displayed much less social learning than HYV wheat. All of these results could still be biased in the presence of spatially correlated profitability shocks. Using detailed information about social interactions Conley and Udry (2003) distinguish geographical neighbors from “information neighbors”, the set of individuals from whom an individual neighbor may learn about agriculture. They show that pineapple farmers in Ghana imitate the choices (of fertilizer quantity) of their information neighbors when these neighbors have a good shock, and move further away from these decisions when they have a bad shock. Conley and Udry try to rule out that this pattern is due to correlated shocks by observing that the choices made on an established crop (maize-cassava intercropping), for which there should be no learning, do not exhibit the same pattern. The ideal experiment to identify social learning is to exogenously affect the choice of technology of a group of farmers and to follow subsequent adoption by themselves and their neighbors, or agricultural contacts. Duflo et al. (2003) performed such an experiment in Western Kenya, where less than 15% of the farmers use fertilizer on their maize crop (the main staple) in any given year despite the official recommendation (based on results from trials in experimental farms), as well as the high returns (in excess 25 For a related model, see Bikhchandani, Hirshleifer and Welch (1992). 26 See Manski (1993) for a discussion of the identification problem in social learning problems. 27 Ellison and Fudenberg (1993) describe “rule of thumb” learning rules where individuals learn from others
only if they are similar.
Ch. 7: Growth Theory through the Lens of Development Economics
517
of 100%) that they estimated. They randomly selected a group of farmers and provided fertilizer and hybrid seeds sufficient for small demonstration plots in these farmers’ fields. Field officers from an NGO working in the area guided the farmers throughout the trial, which was concluded by a debriefing session. In the next season, the adoption of fertilizer by these farmers increased by 17%, compared to adoption by the comparison group. However, there is no evidence of any diffusion: People named by the treatment farmers as people they talk to about agriculture did not adopt fertilizer any more than the contacts of the comparison group. The neighbors of the treatment group actually tended to adopt fertilizer less often, relative to the neighbors of the comparison group. This is not because only experimentation in one’s own field changes someone’s priors: When randomly selected friends were invited to attend the harvest, the debriefing session, and other key periods of the trials, they were as likely to adopt fertilizer as the farmers who participated in the experiment. Rather, it suggests that, spontaneously, information about agriculture is not shared. This points towards another type of externality and source of multiple steady states: When there is very little innovation in a sector, there is no news to exchange, and people do not discuss agriculture. As a result, innovation dies out before spreading, and no innovation survives. Depending on the priors of the individuals, social learning can either decrease or increase investment. In Kenya, Miguel and Kremer (2003) show that random variation in the number of friends of a child who was given the deworming medicine had a negative impact on the propensity of a child to take the medicine. They attribute this to the fact that parents may have initially over-estimated the benefits of the deworming drug. In addition to social learning, there are many other sources of local interactions. First, people imitate each other even when they are not trying to learn, because of fashion or social pressure. Social norms may prevent the adoption of new technologies, because coordinating on a new equilibrium may require many people to change their practices at the same time.28 Second, there are several sources of positive spillovers between industries located close to each other. Silicon Valley-style geographic agglomerations occur in the developing world as well, such as the software industry in Bangalore. Ellison and Glaeser (1997) show that, in the U.S., most industries are indeed more concentrated than they would be if firms decided to place their plants randomly. Only about half of this concentration is explained by the fact that some locations have natural advantages for specific industries [Ellison and Glaeser (1999)]. In addition to the traditional arguments for positive spillovers, such as transport costs (fast telecommunication lines that were installed for the software industry in Bangalore greatly reduced the cost of setting up call centers, for example), intellectual spillovers or labor market pooling, a powerful reason for geographical agglomeration in developing countries is the role of a town’s reputation in the world market. For example, outsiders who want to start working in garment manufacturing come to Tirupur, the small town studied in Banerjee and Munshi (2004), despite their difficulty in finding credit there,
28 See Munshi and Myaux (2002) for an example on the spread of family planning in Bangladesh.
518
A.V. Banerjee and E. Duflo
because this is the place where large American stores come to place orders. There is a sense in which the town has a good reputation, for quality and timeliness of delivery, and everybody who works there benefits from it. Tirole (1996) models “collective reputation”: If many people in a group are known to deliver good quality products, buyers will have high expectations and be willing to trust the sellers to produce more elaborate products, where quality matters. In turn, this will encourage sellers to produce high quality products to avoid being outcast from the group, which will sustain a “high quality-high trust equilibrium”. But if buyers are expected to only ask for basic products in the future, building a reputation for high quality is not useful, and opportunistic sellers will produce low quality in the first period. Knowing this, sellers indeed have the incentive to ask for simple products, and the bad equilibrium persists. In this world, history matters. A collective reputation for low quality is very difficult to reverse, and a collective reputation for high quality is valuable. We should therefore expect groups to try to set up institutions to develop a good collective reputation. There is certainly some indication that this is happening. For example, the association of Indian software firms (NASSCOM) tries to help the firms access quality certifications such as ISO 9001, SEI, or others. Much more work on whether collective reputation matters in practice is, however, clearly needed before we can assess the empirical relevance of these sources of externality. To summarize, externalities can explain very large variations in productivity and investment rates across otherwise similar environments. 4.5. The family: incomplete contracts within and across generations Investment in human capital often pays in the long term, and in many crucial instances must be done by parents on behalf of the child. In this context, the way the decisions are made in the family has a direct impact on investment decisions. In the benchmark neo-classical model [Barro (1974), Becker (1981)], parents value the utility of their children, perhaps at some discounted rate. This world tends to be observationally equivalent to one where an individual maximizes his long run income, and has the same strong convergence properties. However, if parents are not perfectly altruistic, the ability to constrain the repayment of future generations influences investment decisions. Banerjee (2004) studies the short and long run implications of different ways to model the family decision-making process. He shows that incomplete contracting between generations generates potentially large deviations from the very strong convergence property of the Barro–Becker model. Deviations also occur if parents value human capital investment for its own sake (for example, because people like to see their children happy).29 29 Part of the reason why investment in human capital may appear like a preference factor is that individuals
want their offspring to thrive and survive. In the U.S., Case and Paxson (2001) and Case, Lin and McLanahan (2000) find that investment in children is lower when they do not live with their birth mother. Using data from several African countries, Case, Paxson and Ableidinger (2002) find that the gap between the probability of being enrolled in school for orphans and non-orphans can be in part accounted for by the fact that they are less likely to live with at least one parent, and more likely to live with non-relatives.
Ch. 7: Growth Theory through the Lens of Development Economics
519
In particular, even with perfect credit markets, parental wealth will determine how much is invested in human capital. There can be more than one steady state, and there can be inequality in equilibrium. In this world, increases in returns to human capital may not lead to an increase in human capital, if the production of human capital is skillintensive (the increase in the price of teachers may dominate the added incentives to invest in education). Many studies have shown that human capital investment is correlated with family income [see Strauss and Thomas (1995) for references for developing countries]. In general, however, it is difficult to separate out the pure income effect from the effect of an increase in the returns to investing in human capital, differences in the opportunity cost or the direct cost of schooling, and different discount rates. For example, in the Barro–Becker model, families with a lower discount rate will tend to be richer and more likely to invest in education. To avoid this problem, a few studies have focused on exogenous changes in government transfers. For example, Carvalho (2000) shows that an increase in pension income in Brazil led to a decrease in child labor and an increase in school enrollment. Duflo (2003) shows that, in South Africa, girls (though not boys) have better nutritional status (they are taller and heavier) in households where a grandmother is the recipient of a generous old age pension program. This paper also touches on another set of issues. Different members of the family may have different preferences. If education and health were pure investment, and if the members of the household bargained efficiently [as in Lundberg and Pollack (1994, 1996) or the papers reviewed in Bourguignon and Chiappori (1992)], this would not have any impact on education or health decisions. However, if either assumption is violated, it means that not only the size of the income effects, but who gets the income, will affect investment decisions. In the case of the South African pensions, this was clearly the case: Pensions received by men had no impact on the nutritional status of children of either gender. This may come from the fact that women and men value child health differently, or from the fact that the household is not efficient, and a specific individual is more likely to invest in children if the returns are more likely to directly accrue to her. If the household does not bargain efficiently, the consequences extend beyond investment in human capital to all investment decisions. In a Pareto efficient household, production and consumption decisions are separable: The household should choose inputs and investment levels to maximize production, and then bargain over the division of the surplus. This property will be violated if individuals make investment decisions with an eye toward maximizing the share of income that directly accrues to them. Udry (1996) shows that, in Burkina Faso, after controlling for various measures of the productivity of the field (soil quality, exposure, slope, etc.), crop, year, and household fixed effects, yields on plots controlled by women are 20% smaller than yields on men’s plots.30 This does not seem to be due to the fact that women and men have different 30 In Burkina Faso, as in many other African countries, agricultural production is carried out simultaneously
on different plots controlled by different members of the household.
520
A.V. Banerjee and E. Duflo
production functions. Instead, this difference is largely attributed to differences in input intensity: In particular, much less male labor and fertilizer is used on plots controlled by women than on plots controlled by males. The fertilizer result is particularly striking, since there is ample evidence that it has sharply decreasing returns to scale. Udry estimates that the households could increase production by 6% just by reallocating factors of production within the household. Udry explains underinvestment on women’s plots by their fear of being expropriated by their husband if he provides too much labor and inputs. Another reason for inefficient investment may be the fear of being fully taxed by family members once the investment bears fruit. Again, an efficient household would first maximize production. However, the specific claims that a household member (or a neighbor, or a member of the extended family) can make on someone’s income stream may lead to inefficient investment. Consider, for example, a situation where individuals have the right to make emergency claims on the income or savings of others in their group (for example, if someone is sick and has no money to pay for the doctors, others in his extended family have an obligation to pay the doctor). Consider a savings opportunity that will increase income by a large amount in the future (for example, saving money after harvest to be able to buy fertilizer at the time of planting). If everybody could commit not to exercise their claim during the period where the income needs to be saved, the money should be saved, and the proceeds eventually distributed to those who have a claim on it, and everybody would be better off. However, if no such commitment is possible, the individual who earned the income knows that it is likely that, should he choose to save enough for fertilizer, a claim will be exercised in the period during which the money needs to be held. He is then better off spending the money right away: Even if individuals are rational and have a low discount rate, as a group they will behave as “hyperbolic discounters”, who discount the immediate future relative to today more than future periods relative to each other [Laibson (1991)]. The level of investment will be low in the absence of savings opportunities offering some commitment to household members. The fact that investments are often decided within a family, rather than by a single individual, or that the proceeds of the investment will be shared among a set of people who have not necessarily supported the cost of the investment therefore greatly complicates the incentive to invest. This may, once again, explain why some potential investments with high marginal product are not taken advantage of. It is worth noting that the lack of credit and insurance in poor countries makes these problems particularly acute there. For example, the lack of credit markets means that investment decisions are taken within the families – e.g., women cannot borrow to get the optimal amount of fertilizer on their plot – and the lack of insurance plays an important role in justifying the norms on family solidarity that seem to be hindering productive investment. 4.6. Behavioral issues Individuals in the developing world appear not only to be credit constrained, but also to be savings constrained. Aportela (1998) shows that when the Mexican Savings In-
Ch. 7: Growth Theory through the Lens of Development Economics
521
stitution “Pahnal” (Patronoto del Ahorro Nacional) expanded its number of branches through post offices in poor areas and introduced new savings instruments in the 1990s, household’s savings rates increased by 3% to 5% in areas where the expansion took place. The largest increase occurred for low income households. When an individual (or his household) has time-inconsistent preferences, formal savings instruments may increase savings rates even when they offer very low returns (even compared to holding onto cash), because they offer a commitment mechanism. Microcredit programs may also be understood as programs helping individuals to commit to regular reimbursements. This is particularly clear for programs, like the FINCA program in Latin America, which require that their clients maintain a positive savings balance even when they borrow.31 Duflo et al. (2003) provide direct evidence that there is an unmet demand for commitment savings opportunities among Kenyan farmers, and that investment in fertilizer increases when households have access to this opportunity. In several successive seasons, they offered farmers the option to purchase a voucher for fertilizer right after harvest (when farmers are relatively well off). The vouchers could be redeemed for fertilizer at the time when it is necessary to plant it. The take up of this program was quite high: 15% of the farmers took up the program the first time it was tried with farmers who had never encountered the NGO before. Net adoption of fertilizer increased in this group. The program was then offered to some of the farmers who had participated in the pilot program mentioned above (and thus had the opportunity to test the fertilizer for themselves, and trusted the NGO), and in this group, the take up was 80%. There is also direct evidence of the difficulty for farmers to hold on to cash: In other experiments, when farmers were given a few days before they could purchase the voucher, the take up fell by more than 50%. When they were offered the option of having the fertilizer delivered at home at the time they actually needed it (and to pay for it then), none of the farmers who had initially signed up for the program had the money to pay for the fertilizer when it was delivered. This area of research is quite recent, and wide open. Many questions need answering, and the area of applicability is wide. For example, what is the best way to increase parents’ willingness to invest in deworming drugs? Why don’t all parents sign the authorization form which will grant free access to deworming to their children [Miguel and Kremer (2003)]? Is it a rational decision or is it procrastination? Why does the take up of the deworming drug fall so rapidly when a small cost-sharing fee is introduced [Miguel and Kremer (2003)]? Understanding the psychological factors that constrain investment decisions, and the role that social norms play in disciplining individuals, but also potentially in limiting their options, is an important area for future research. Several randomized evaluations are trying to make progress in this area. They are addressing 31 Karlan (2003) argues that simultaneous borrowing and savings by many clients in these institutions can
be explained by the value to the small business owner of the fixed repayment schedule as a discipline device. Gugerty (2000) and Anderson and Baland (2002) interpret rotating credit and savings (ROSCAs) institutions in this light.
522
A.V. Banerjee and E. Duflo
questions as diverse as: What is the role of marketing factors in the access of poor people to loans in South Africa [Bertrand et al. (2004)]? Do poor people take advantage of savings products with commitment options in the Philippines [Ashraf, Karlan and Yin (2004)]? What prevents people from doing a small action that would lead them to a high return [Duflo et al. (2003)]? What factors (deadline, framing, etc.) make it more likely they will take an action [Bertrand et al. (2004)]? A defining characteristic of these projects is that they do not involve laboratory experiments: Like the research on fertilizer in Kenya, they set up real programmes which are likely to increase poor people’s investment and improve welfare if they indeed deviate from perfect rationality in the way the psychological literature suggests. In order to be fruitful, this agenda will need to avoid simply transplanting to developing countries some of the insights developed by observing behaviors in rich countries. Being poor almost certainly affects the way people think and decide. Decisions, after all, are based not on actual returns but on what people perceive the returns to be, and these perceptions may very well be colored by their life experience. Also, when choices involve the subsistence of one’s family, trade-offs are distorted in different ways than when the question is how much money one will enjoy at retirement. Pressure by extended family members or neighbors is also stronger when they are at risk of starvation. It is also plausible that decision-making is influenced by stress. What is needed is a theory of how poverty influences decision making, not only by affecting the constraints, but by changing the decision making process itself.32 That theory can then guide a new round of empirical research, both observational and experimental.
5. Calibrating the impact of the misallocation of capital In this long list of potentially distorting factors there are some, like government failures or credit market failures, that most people find a priori plausible, and others, such as intra-family inefficiencies or learning externalities, that are more contentious, and yet others, like the behavioral factors, that have not yet been widely studied. However, even where the prima facie evidence is the strongest, we cannot automatically conclude that the particular distortion has resulted in a significant loss in productivity. To get a sense of the potential productivity loss, we return to the Indo-U.S. comparison. Taking as given the stock of capital in India and the U.S. today, any of the multiple distortions listed above could have affected productivity in two different ways: First, there may be across-the-board inefficiency, because everyone could have chosen the wrong technology or the wrong product mix. Second, capital may be misallocated across firms: There may be differences in productivity across firms, either because of differences in scale, or because of differences in technology or because some entrepreneurs are more skilled than others, and the distribution of capital across these firms may be sub-optimal, in the sense that the most productive firms are too small. 32 See Ray (2003) for a very nice attempt to start in this direction.
Ch. 7: Growth Theory through the Lens of Development Economics
523
Here we have chosen to emphasize this latter source of inefficiency, motivated in part by the evidence, discussed above, that tells us that there are enormous differences in productivity across firms. We take no stance on how such an inefficient allocation of capital came about, nor on why the firms do not make the right choices, either of scale or technologies. Lack of access to credit is, of course, a potential explanation for both, but it could equally be explained by lack of insurance, the fear of confiscation by the government, or the gap between real and perceived returns. The goal of this section is to set-up and calibrate a simple model, to investigate whether the misallocation of capital across firms within a country can explain the aggregate puzzles we started from: the low output-per-worker in developing countries, given the level of capital, and the low marginal product of capital, given the outputper-worker. We do not claim that we necessarily have the right explanation; there are simply too many degrees of freedom in this kind of exercise. Nevertheless we feel that the exercise has some value, not least because it gives us some reason to believe that we have not been entirely misguided in emphasizing the role of misallocation. Moreover, as we will see, it does rule on some kinds of misallocation stories in favor of others. We begin with a model where the misallocation only affects the scale of production, because all the firms share the same technology. Scale obviously does not matter where there are constant returns to scale, so we need to turn to a model where there are diminishing returns at the firm level.33 We will show that, with realistic assumptions about the relative firm size in India and the U.S., this model cannot go very far in explaining the aggregate facts. We then turn to a model where a better technology can be purchased for a fixed cost. We show that this model, coupled with the misallocation of capital, will help generate the aggregate facts, with realistic assumptions about the distribution of firm sizes. 5.1. A model with diminishing returns • Model set-up Consider a model where there is a single technology that exhibits diminishing returns at the firm level, say, Y = ALγ K α , with γ < 1−α. Also, we will assume that the economy has a fixed number of firms: Without that assumption, everyone will set up multiple minuscule firms, thereby eliminating the diminishing returns effect. To justify this we make the standard assumption that the economy has a fixed number of entrepreneurs and each firm needs an entrepreneur. Under these assumptions every firm would invest the same amount when markets function perfectly, but when different firms are of different sizes, the marginal product would vary across the firms and efficiency would suffer. The question is whether these effects are large enough to help us explain what we see in the data. Given that we are 33 The obvious alternative – increasing returns at the firm level – will clearly not fit the basic fact that there
is more than one firm in the U.S., or that the marginal product of capital is higher in India than in the U.S.
524
A.V. Banerjee and E. Duflo
still within the class of Cobb–Douglas models, we know from Section 3 that we cannot get both the right ratio for output-per-worker and the right value of the ratio of marginal products; we therefore focus on the output-per-worker. We start with a population of firms indexed by i, and that firms face a limit on how much they can borrow, so that for firm i, K K(i). The demand for labor from a α
1
firm that invests K(i), is given by [ Aγ K(i) ] 1−γ . We assume a perfect labor market, so w that given the level of capital, labor is efficiently allocated across firms. Labor market clearing then requires that
w = Aγ
1
[K(i)α ] 1−γ dG(i) L
1−γ ,
where G(i) represents the distribution of i and L is labor supply per firm. Since wages are a fraction γ of output-per-worker, it follows that output-per-worker will be 1
[K(i)α ] 1−γ dG(i) 1−γ . A L
Consider an economy where, for any of the reasons we have outlined above, some firms have access to more capital than others. In particular, assume that in equilibrium a fraction λ of firms get to invest an amount K 1 and the rest get to invest K 2 > K 1 .34 This would clearly explain why the marginal product of capital varies within the same economy. We would also expect that this inefficiency in the allocation of capital would lower productivity relative to the case where capital was optimally allocated. To get at the magnitude of the efficiency loss, note that output-per-worker in this economy will be:
λ(K 1 )α/(1−γ ) + (1 − λ)(K 2 )α/(1−γ ) 1−γ A . L We compare this economy with another which has a TFP of A , a labor force L and a capital stock K , which is, in contrast with the other economy, allocated optimally across firms. To say something about productivity we also need to say how many firms there are in this economy. Let us start by assuming that the number of firms is the same. Then the ratio of output-per-worker in our first economy to that in the second is: K K α 1−α−γ A/A L /L L L α/(1−γ ) α/(1−γ ) 1−γ × λ K 1 /K + (1 − λ) K 2 /K . (6)
34 Since all firms face the same technology and there are diminishing returns to scale, this would not happen in the absence of these imperfections (all the firms should invest the same amount, λK 1 + (1 − λ)K 2 = K).
Ch. 7: Growth Theory through the Lens of Development Economics
525
We already noted that for the India–U.S. comparison, the ratio L/L
K L
K L
is about 1 : 18.
The same source (the Penn World Tables) tells us that is about 2.7. What are reasonable values of α and γ ? For 1 − α − γ , which is the share of pure profits in the economy, we assume 20%, which is what Jovanovic and Rousseau (2003) find for the U.S. This is presumably counted as capital income, so we keep γ = 0.6 and set α = 0.2. First consider the case where λ = 1, so that capital is efficiently allocated in both countries. Then the productivity ratio ought to be (A/A )( K K )α (L /L)1−α−γ : AsL L suming that 2A = A , as before, because of the human capital differences across these 1 0.2 1 0.2 economies, the ratio works out to be 12 ( 18 ) ( 2.7 ) = 23%. Recall from Equation (5) that the model with constant returns predicted that output should be 6.35 higher in the U.S., or, equivalently that output per capita in India should be 15.7% the U.S. level. The 23% predicted by the current model is, of course, even further from the 9% we find in the data. The reason why this model does worse is because the production function is more concave: The concavity penalizes the U.S., which has more capital relative to India. • What if capital is misallocated? To bring in the effects of misallocating capital, we need to determine the size of the gap between K 2 and K 1 that we can reasonably assume. One way to calibrate these numbers is to make use of the estimate of Banerjee and Duflo (2004) that in India, there are firms where the marginal product of capital seems to be close to 100%. On the other hand, some seem to have access to capital at 9% or so, and therefore may well have a marginal product reasonably close to 9% [Timberg and Aiyar (1984)]. From the production function, we know that if we assume that K 1 corresponds to the firm with a marginal product of 100%, while K 2 is the firm with the marginal product of 9%, then α −1 9 2 (K 2 /K 1 ) 1−γ = 100 or K 2 /K 1 = ( 100 9 ) = 123. We can now evaluate the ratio of output-per-worker in the two economies for any given value of λ, the fraction for firms with capital stock K 1 . To pin down λ, we use the fact that the average of the marginal product in India seems to be somewhere in the range of 22%. In our model, under the assumption that the marginal dollar is allocated between small firms and large firms in the same proportion as the average dollar, the average marginal product of capital is given by: λ (1 − λ)123 100 + 9. λ + 123(1 − λ) λ + 123(1 − λ) Since this is equal to 22% we have that λ = 0.95. We can now compute the extent of productivity loss due to the misallocation. From Equation (6), this is given by the expression [λ(K 1 /K)α/(1−γ ) + (1 − λ)(K 2 /K)α/(1−γ ) ]1−γ . Under the assumed values, it is approximately 0.8. In other words, the misallocation brings the productivity ratio we expect to see between India and the U.S. down from 23% to about 18%. Relative to the neo-classical model we started from (which generates an output per worker in India of 15.7% of the U.S. level), moving to this model therefore does not
526
A.V. Banerjee and E. Duflo Table 3 Distribution of firm size (Annual Survey of Industry, 2000) 95-5 ratio
median-5 ratio
mean-5 ratio
5th percentile
1007 786 1392 22300 3093 1581 235 2639 1700
95 150 90 440 31 104 10 122 29
216 285 620 5423 1292 410 53 683 504
61466 29899 5681 12870 14159 22461 37075 21825 84103
Manufacture of pasteurized milk Flour milling Rice milling Cotton spinning Cotton weaving Textile garment manufacture Curing raw hides and skins Manufacture of footwear Manufacture of car parts
help close the productivity gap between India and the U.S. The problem is, once again, that the additional productivity gap that the misallocation generates is more than compensated for by the effect of making the production function concave while keeping the number of firms fixed. • Can we do better than this? We could try to make the misallocation problem worse in India by making the big firms bigger and the small firms smaller. However the industry structure we started with was already rather extreme. Table 3 lists, for nine of the largest industries in India (where industry is defined at 3 digit level) outside of agriculture, known for having a substantial presence of small enterprises, some measures of variation in firm sizes (where size is measured by the net fixed capital in year 2000) from the Annual Survey of Industries (ASI). In the industry described by our model, the large firm in our model is 123 times the size of the small firm whereas in the ASI data, even the 95th percentile firm in the 1 median industry is no more than 72 times the 25th percentile firm. The firm that is 123 times the 95th percentile firm in the median industry is around the 20th percentile in the size distribution. More than 50% of the capital stock in the Indian economy is in firms that are bigger than the “small” firm and smaller than the “large” firms as we defined them here. Realism therefore requires that we move weight away from the extremes, but this will not help us fit the data, since it dampens the concavity effect that is at the heart of our argument. Another problem with making the big firms bigger is that the big firms in our data are already too big, relative to their American counterparts: Since K 2 = 123K 1 , K I = [0.955 + 123(0.045)]K 1 = 6.5K 1 and K 2 ≈ 19K I . Now since ( K )I ( K )U is about L
L
1/18 and LI /LU is about 2.7, K I /K U = 0.15. Therefore K 2 /K U = 2.85. The biggest U.S. firms in our model are a third of the biggest Indian firms, whereas it is typically assumed that big U.S. firms are, if anything, bigger than big Indian firms.
Ch. 7: Growth Theory through the Lens of Development Economics
527
Finally note that in our model, the ratio of the 95th percentile firm to the 5th percentile firm in the median industry is approximately 1600 : 1.35 Given the production function, we know that the marginal return on capital in the two firms should differ by a factor of 16001/2 = 40 : 1. If the biggest firms pay about 9% for their capital, the smaller firms must have a marginal product that exceeds 360%, which seems implausible. Making the gap between small and large firms even larger would clearly exacerbate this problem. Another possible way to augment the productivity gap is to give up the assumption that the two economies have same number of firms. Suppose the U.S. had λ > 1 times as many firms as India: Then the labor productivity ratio computed above would have to be divided by λ1−α−γ . If λ were equal to 32, the ratio of labor productivity in India to that in the U.S. would be 9%, which is what we find in the data. Of course, increasing the number of the firms in the U.S. will tend to make the average firm in the U.S. smaller: Even with the same number of firms in the two countries, the fact that the biggest firms in India have about 18 times the average capital stock means that they are about 3 times the size of U.S. firms, which seems implausible. If there are 32 times as many firms in the U.S., the average U.S. firm would be about a 1/100th of the biggest Indian firm, close to 25% in the Indian size distribution. This seems entirely counterfactual. • Taking stock To sum up, moving to this more sophisticated model does not help us fit the macro facts better. It obviously does suggest a simple theory of the cross-sectional variation in returns to capital, which is entirely absent from the model with constant returns, but ends up throwing up a number of other problems that a theory of this type will need to deal with. In particular, a successful explanation has to be consistent with the fact that the firm size distribution in India has a large part of its weight near the mean/median; that even the biggest Indian firms are not larger than the bigger U.S. firms in the same industry; and that the marginal product of capital in the average small Indian firm, while large (even 100%) is probably not 300% or more. The next section introduces an alternative model where firms differ both in scale and in technology, but still retains the assumption that there is no inherent difference between these alternative investors. 5.2. A model with fixed costs • Model set up Consider a world where setting up requires a fixed start-up cost in addition to an entrepreneur, but once these are in place, capital and labor get combined as in a standard Cobb–Douglas with diminishing returns. This fixed cost could come from many sources: Machines come in certain discrete sizes and even the smallest machine may be 35 The median industry is the Textile Garment Manufacturing industry.
528
A.V. Banerjee and E. Duflo
expensive from someone’s point of view. Buildings, likewise, are somewhat indivisible, at least by the time we come down to a single room or less. Marketing and building a reputation may also requires an indivisible up-front investment – Banerjee and Duflo (2000) describe the costs that a new firm in the customized software industry has to pay in terms of harsh contractual terms, until it has a secured reputation. Turning to investment in human capital, it also appears that the first five years or so of education may have much lower returns than the next few years, which in effect makes the first few years of education a fixed cost.36 Finally, as emphasized by Banerjee (2003) the fixed cost may be in the financial contracting that the firm has to go through – starting loans are often expensive because the lender cannot trust the borrower with a big loan and when the loan is small, the fixed costs of setting up the contract loom large. Formally, we assume a production function y = A(K − K)α Lγ . Since we continue to assume that the firm can buy as much labor as it wants, the production function can be rewritten as:
γ 1 α γ 1−γ 1−γ A (7) [K − K] 1−γ . w We continue to assume that γ + α < 1, so that there are diminishing returns. The average cost function in this world has the classic Marshallian shape: Average costs go down first as the fixed costs get amortized over more and more output and then start to rise again. The optimal scale of production is given by the equality of the marginal and average product of capital, which reduces to: K=K
1−γ . 1−γ −α
We allow firms the option of choosing between alternative technologies. Assume that there are three alternative technologies available, characterized by three different levels of the fixed cost, K 1 , K 2 and K 3 , three differing levels of labor and capital intensity, {(α1 , γ1 ), (α2 , γ2 ), (α3 , γ3 )} and three correspondingly different levels of productivity, A1 , A2 and A3 . We make the usual assumption that a higher cost buys a higher level of TFP, i.e., that K 1 K 2 K 3 and A1 A2 A3 . Compared to a Cobb–Douglas model with diminishing returns, this formulation has a number of advantages. First, it allows firms to have large differences in size without necessarily large differences in the marginal product of capital, since they could be using different technologies. The fact that there are firms in the same industry operating at very different scales posed a problem for the model with diminishing returns because the implied variation in the marginal product of capital seems implausibly large. Second, the fact that a lot of the capital stock in India is in firms that are close to the mean, was
36 All the estimates (14) we could find of Mincerian returns at different levels of education suggest that, in
developing countries, the marginal benefit of a year of education increases with the level of education (in the U.S., it appears to be very flat). Schultz (2004) finds the same result in his study of six African countries.
Ch. 7: Growth Theory through the Lens of Development Economics
529
a problem when we had global diminishing returns, because with diminishing returns, firms close to the mean are at the optimal scale. In our present model, the right scale for Indian firms is actually much bigger than the mean. A part of the inefficiency comes precisely from the fact that there are many firms that are concentrated near the mean. Third, as noted above, this model generates a unique optimal scale of production, which would provide a reason why the biggest (and most productive) Indian and U.S. firms could be more or less the same size. Fourth, because efficient firms tend to be quite large, it is easy to see why India, with its multitude of firms that are too small, will be inefficient relative to the U.S., where all firms are at the right scale. Fifth, the fact that production requires a fixed cost helps explain why, despite the diminishing returns from technology, we do not see people setting up a very large number of very small firms, thereby completely eliminating the diminishing returns effect. In this case, we can let the number of firms be determined by what people are willing to invest, in combination with what we know about the fixed costs (actually as noted below, we cheat slightly on this point, but only because it simplifies the calculations). Sixth, the fact that we allow the number of firms to be determined endogenously means that there are less overall diminishing returns. When we compare the U.S. and India, this helps explain why the productivity gap is so large and why interest rates are not lower in the U.S. Finally making this assumption alters the nature of the link between the marginal product of capital and its average product. With a Cobb–Douglas, the ratio of the average product is always proportional to the marginal product. Here, the average product starts lower than the marginal product but grows faster and eventually becomes larger than the marginal product. In other words, as firm size goes up the ratio of the marginal product of capital to its average product goes down, at least initially. This would suggest that the ratio of the average products of capital in India and the U.S. should be less than the ratio of the marginal products, and indeed we find that while output-per-worker is 11 times larger in the U.S., capital-per-worker is 18 times as large, implying an average product ratio of about 1.6 : 1, as against the 2.5 : 1 ratio of marginal products delivered by the standard Cobb–Douglas model. This is clearly an a priori advantage of this formulation, since, as we noted in Section 3, the proportionality between the average product and the marginal product prevents any model based on a Cobb–Douglas production function to fit these facts. Interestingly, this model brings together elements – market imperfections and some increasing returns – that are also being invoked by recent work in new growth theory [Aghion et al. (2004), for example] for the same purpose, namely to explain the lack of convergence. However, the increasing returns and the credit constraints here are at the level of the firm, whereas in the aggregative growth literature they are at the level of the economy. Indeed, if there is no misallocation and no lack of people to start new firms, the aggregate production function generated by our economy exhibits constant returns in labor and capital: Indeed it has exactly the form that Lucas started with – Y = AK α L1−α . In order to impose restrictions on the parameters of the model, we make use of the industry data described in Table 3. We describe the representative Indian industry by
530
A.V. Banerjee and E. Duflo
a 3-point distribution of firms sizes, with fractions λ1 , λ2 , and λ3 at K1 , K2 and K3 . The first group of firms is made of the bottom 10% of the distribution of firms, and we assigned to them the size of the firm at the 5th percentile of the actual size distribution in the data. Likewise, we assume that the top 10% of all firms are in the group of “large firm”, and that their size is that of the firm at the 95th percentile of the firm size distribution.37 The rest we assign to the middle category, whose size we set at the mean for the distribution. We assume that the largest firm is 1,600 times as big as the smallest firm, which is roughly the median value of these ratios across these nine industries in our data. These parameter values imply that the mean firm size in the industry will be 800 times as large as the 5th percentile firm, which is higher than the mean in the median industry in our data (500 times), but well within the existing range in the data. Once again we are interested in the within-economy variation in returns to capital. We therefore assume, as before, that the small firms have a marginal product of 100% while the medium sized firms have a marginal product of just 9%. The more unorthodox assumption is that the large firms also have a marginal product of 100%. While clearly somewhat artificial, this is meant to capture the idea that the best technology is expensive and only the biggest firms in India can afford to be at the cutting edge, an idea that is very much in the spirit of the McKinsey Global Institute’s study of a number of specific Indian industries. However, they are still relatively small and therefore the marginal returns on an extra dollar of investment are very high. The rest of the firms use cheaper (i.e., lower K) but less effective technologies. In particular, the small firms are simply too small (which explains their high marginal product), and the middle category consists of firms that have exhausted the potential of the mediocre technology that they can afford but are too small to make use of the ideal technology. How plausible is our assumption about industry structure? The average capital stock of the 95th percentile firms in the median industry was Rs. 36 million, which puts them at a size just above the category of firms that are the focus of Banerjee and Duflo (2004). The point of that paper was that a subset of these firms (the firms that attracted the extra credit after the policy change) had marginal returns on capital of close to 100%. Therefore it is not absurd to assume that the large firms in our model economy have very high returns. Once we accept the idea that some large firms are very productive, given that the average marginal product is probably close to 22%, it is obviously very likely that there are many smaller firms that have a lower marginal product than the largest firms. Indeed, when we calculate the average marginal product based on our assumptions, under the premise that the marginal dollar is distributed across the three size categories in the ratio of their share in the capital stock, the average marginal product turns out to be about 27%. Even with this long list of assumptions, we do not have enough information to compute output-per-worker in our model economy – there are several remaining degrees 37 We pick the 5th and the 95th percentile to make the difference in the returns to capital between the biggest
and smallest firms as large as possible.
Ch. 7: Growth Theory through the Lens of Development Economics
531
of freedom. First, we need to choose units: Our assumption, which simplifies calculations, is that capital is measured in multiples of the small firm. Finally, we assume that K 1 = 0, K 2 = 100, and K 3 = 800. The assumption that K 3 = 800, implies that the biggest Indian firms (which have 1,600 units of capital) are operating at the bottom of 1−γ 38 the average cost curve – given by K 1−γ −α . • Results: Output-per-worker and average marginal product of capital Under these assumptions, we can use the assumed marginal products to solve for A1 , A2 and A3 . According to these calculations, TFP in the medium firms is about 1.4 times bigger than that in the small firms, and that in the large firms is about 2.7 times that in the medium firms. Nevertheless, given the assumed limits on how much they can invest, each category of firms is optimizing by choosing its current technology. However, large gains in productivity are obviously possible if the economy can reallocate its capital so that all the firms adopt the most productive technology. To see how large this gain may be, we do another India–U.S. comparison. Once again we assume that the U.S. takes full advantage of the available technology. In other words, every firm in the U.S. operates the best technology at the optimal scale, i.e., each of these firms operates technology 3 and has 1,600 units of capital. It is easily checked that the aggregate production function for the U.S. implied by these assumptions if of the form Y = AK 0.4 L0.6 . If India also operated at full efficiency (i.e., the production function is the same as in the U.S.) we already know from our calculations based on the Lucas model that the ratio of output-per-worker in the two countries would be 6.35 : 1. Our key assumption is that the distribution of firm sizes in India, by contrast with the U.S., includes a large fraction of firms that neither use the best technology nor operate at the optimal scale. The implicit premise is that in the U.S. there are enough people who are able and willing to invest 1,600 units if there is any money to be made, but this is not true in India because of borrowing constraints or other reasons. A series of straightforward calculations gives the expression for the ratio of outputper-worker, which is also the ratio of wages in the two economies: 1 1 wI 1−γ yI 1−γ = yU wU 1 α 1 α N I LU λ1 (AI ) 1−γ (K1 − K 1 ) 1−γ + λ2 (AI A2 /A1 ) 1−γ (K2 − K 2 ) 1−γ = N U LI 1 α 1 α (AU A3 /A1 ) 1−γ (K3 − K 3 ) 1−γ , + λ3 (AI A3 /A1 ) 1−γ (K3 − K 3 ) 1−γ where NI and NU are the numbers of firms in India and the U.S., and AI and AU represent the base levels of TFP. The only reason that AI = AU is, as before, that 38 This is where we cheat, since with decreasing returns to scale, there could again be an infinity of very
small firms, so that all the firms should be in the small group. We can prevent that if we assume that the smallest feasible firm size is actually ε greater than zero, and only a certain number of entrepreneurs are able (or willing) to invest at least ε.
532
A.V. Banerjee and E. Duflo
the human capital levels vary. We continue to assume that AU = 2AI . NI /NU can be computed from the fact that the total demand for capital from these firms must exhaust the supply of capital: i.e., KI NI [λ1 K1 + λ2 K2 + λ3 K3 ] = , KU NU K3 which, given the assumed parameter values, implies that NI /NU = 0.3, which can then be used to calculate yI /yU , which turns out to be almost exactly 1/10, not too far from the 1/11 that we found in the data. We can also derive, as before, what this model tells us about the marginal product of capital in the U.S. Using the expression derived in the previous subsection, it is easily shown that the ratio of the marginal product of capital in the U.S. to that in the biggest I
1
γ
and best Indian firms will be given by ( AAU ) 1−γ (wU /wI ) 1−γ ,39 which turns out to be 6.45. Given that the biggest Indian firms have a marginal product of 100%, the average U.S. firms should have a marginal product of 100/6.45 = 15.5%. This is obviously higher than the average stock market return but hardly beyond the reasonable range. • Distribution of firm sizes The most obvious advantage of the fixed cost approach is that we do not obtain the unreasonably large gap in the marginal products of capital between small and large firms within the same economy that came out of the previous model. This underscores the importance of using evidence on cross-sectional differences within an economy to assess the validity of alternative models. Finally, the success of this model in explaining the productivity gap depends, as in the case of the previous model, heavily on the assumption that the U.S. has many more firms than India. However, while in that model we needed the U.S. to have 32 times as many firms as in India to fit the observed productivity gap, here we are doing very well with the U.S. having 3 13 times as many. How reasonable is the assumption that the U.S. has more firms than India? This is not an easy question to answer, mainly because we have no clear sense of what should count as a firm: Both economies have enormous numbers of tiny firms that reflect what people do on the side. In India these “firms” are concentrated in a few sectors, such as retailing or the collection of leaves, wood or waste products, which require little or no skills and can be done on part-time basis. In the U.S., the equivalent would be the numerous ways in which you end up owning a small business for tax purposes, such as part-time consulting, renting out part of your home, part-time telemarketing, etc. It is not clear which of these should count as legitimate firms from the point of view of our model and which of these should not. A way to restate the same point is that by focusing on the median industry in the ASI data, we have effectively ignored the industries (like the ones listed above) which 39 The fact that the biggest firms in India are the same size as any U.S. firm obviously simplifies the calcula-
tion.
Ch. 7: Growth Theory through the Lens of Development Economics
533
attract all those in India who have nowhere better to go. While there are only a few such industries, they are enormous, and quite unlike the rest of the industries: Among the industries listed in the table above, cotton spinning is probably most like what one of these industries looks like, and it is apparent that it is quite different from the rest – there are many more tiny firms. However this is not a problem for what we are doing here. Starting with the Indian firm size distribution assumed in the above exercise, we could simply add many more of the smallest firms to the Indian firms size distribution, until we get to the point where India has the same number of firms as the U.S. Since we have increased the number of firms in India by 3 13 times while keeping the number of large firms (firms with 1,600 units of capital) constant, the share of these firms goes down to 3% (from 10%). These two versions of the Indian economy are reasonably similar, because the smallest firms do not add up to a large amount of capital, but it is obvious that this economy will be less productive than the previous one (since inefficient small firms will have a larger share of total capital), and hence we will actually get somewhat closer to the 11 : 1 productivity ratio that we were shooting for. • Why doesn’t capital flow to India? Finally we subject this model to an additional test: The fact that in our model there are firms in India with returns in the neighborhood of 100% would suggest that there are many unexploited opportunities. We have already argued that there are many reasons why a U.S. bank could not just lend to an Indian firm, and thereby benefit from these opportunities. Nor is it easy for an American to borrow money in the U.S. and set up a firm in India: Once he is in India he may be beyond the reach of U.S. law and for that reason alone, lenders will shy away from him. What is much more plausible, however, is that a U.S. entrepreneur moves to India to invest his money in these opportunities. The question is why this does not happen more often. There are some obvious answers to this question: If the reason why these opportunities have not already been taken is the lack of secure property rights in India, there is no reason why foreigners would be particularly keen to invest in India. On the other hand, if the problem is that Indians do not have the capital or that they fear the risk exposure or that they are simply unaware of the opportunity, to take some plausible alternatives, a well-diversified wealthy U.S. investor may well be attracted to move to India and start a firm. How much money would such an investor make? To answer this we start by observing from (7) that the production function in the largest Indian firms can be written as 1
γ
γ 1−γ C(K − 800)1/2 , where C = A31−γ [ w ] . Of this, a fraction 3/5 goes to wages. Profits 2 are therefore given by 5 C(K − 800)1/2 . Since this firm has 1,600 units of capital, and the marginal product of capital in this firm was assumed to be 100%, it follows that
1 C(800)−1/2 = 1, 5
534
A.V. Banerjee and E. Duflo
or C = 141.42. The opportunity cost of capital for a U.S. investor is 9%. The optimal investment in this Indian firm for a U.S. investor who can invest as much as he wants will be given by the solution to (141.42)(0.2)(K − 800)−1/2 = 0.09. This tell us that the optimal investment is K = 99564. The total after-wage income generated by the firm is (0.4)(141.42)(99564 − 800)1/2 = 17777. This is in units of the smallest firm. We know that the biggest firms in our model are 1,600 times as large as the smallest firms and from the table above, such firms have Rs. 36 million worth of capital in the median industry. The smallest firm therefore has Rs. 22,500 worth of capital, which implies that the U.S. investor will earn 17777(22500) = Rs. 400 million on his investment of (99564)(22500) = Rs. 2.24 billion. This is a net gain of about Rs. 200 million, or about 4 million dollars. Is this a large enough gain to tempt someone to leave his home and family and settle in India? For someone with an average income, obviously. But no one with an average income has 50 million dollars of his own that he is willing to put into a single project in India. Anyone who is willing to do it has to be very rich indeed – he must have $50 million several times over. How many people are so wealthy that they are willing to give up their life in the U.S. for an extra $4 million per year? In other words, while the model developed in this section generates very large productivity losses, it does not offer any one person the possibility of arbitraging these unexploited opportunities to become enormously rich. This is because diminishing returns set in quite fast. 5.2.1. Taking stock We started by describing some of the major puzzles left unanswered by the neo-classical model, and in particular the productivity gap between rich and poor countries. The coexistence of high and low returns to investment opportunities, together with the low average marginal product of capital, suggested that some of the answer might lie in the misallocation of capital. The microeconomic evidence indeed suggests that there are some sources of misallocation of capital, including credit constraints, institutional failures, and others. In this section, we have seen that, combined with multiple technological options and a fixed cost of upgrading to better technologies, a model based on misallocation of capital does quite well in terms of explaining the productivity gap. The value of the marginal productivity of capital in the U.S. predicted by this model is only marginally too high, and the degree of variation in the marginal product of capital within a single economy that the model requires is not implausibly large. Of course the model does make unrealistic assumptions – there is, for example, surely some amount of inefficiency in the U.S., and some U.S. firms are surely more productive
Ch. 7: Growth Theory through the Lens of Development Economics
535
than others. On the other hand, we have also ignored many reasons why Indian firms may be less efficient than they are in our model. For example, our current model assumes that only 10% of the firms, who use less than 1% of the capital stock and produce less than 1% of the output, use the least efficient technology whereas the MGI report on the apparel sector tells us that almost 55% of the output of the sector is produced by tailors who still use primitive technology. We also assumed that 10% of Indian firms are as productive as the best U.S. firms. Clearly that fraction could be smaller. We also assumed that everyone is equally competent. In the real world, imperfect credit markets, for example, drive down the opportunity cost of capital and this encourages incompetent producers to stay in business. In the model, we assume that all large firms earn high returns but in reality there are probably some large firms that have much lower productivity (anywhere down to 9% per year would be consistent with our model). This too will drive down productivity. In a recent paper, Caselli and Gennaioli (2002) try to calibrate the impact of this factor in the context of a dynamic model with credit constraints. They show that in steady state this can generate productivity losses of 20% or so. We will argue in the next section that this severely understates the potential productivity gap starting from an arbitrary allocation of capital. 6. Towards a non-aggregative growth theory 6.1. An illustration The presumption of neo-classical growth theory was that being a citizen of a poor country gives one access to many exciting investment opportunities, which eventually lead on to convergence. The point of the previous section was to argue that most citizens of poor countries are not in a position to enjoy most of these opportunities, either because markets do not do what they ought to or the government does what it ought not to, or because people find it psychologically difficult to do what is expected of them. What can we say about the long-run evolution of an economy where there are rewarding opportunities that are not necessarily exploited? In this section we will explore this question under the assumption that the only source of inefficiency in this economy comes from limited access to credit. The goal is to illustrate what non-aggregative growth theory might look like, rather than to suggest an alternative canonical model. The model we have in mind is as follows: There are individual production functions associated with every participant in this economy that are assumed to be identical and a function of capital alone (F (K)) but otherwise quite general. In particular, we do assume that they are concave. Individuals maximize an intertemporal utility function of the form: ∞ δ t U (Ct ), 0 < δ < 1, t=0
U (Ct ) =
c1−φ , 1−φ
φ > 0.
536
A.V. Banerjee and E. Duflo
People are forward-looking and at each point of time they choose consumption and savings to maximize lifetime utility. However, the maximum amount they can borrow is linear and increasing in their wealth and decreasing in the current interest rate: An individual with wealth w can borrow up to λ(rt )w. Credit comes from other members of the same economy and the interest rate clears the credit market. We do not assume that everyone starts with the same wealth, but rather that at each point of time there is a distribution of wealth that evolves over time. This model is a straightforward generalization of the standard growth model. What it tells us about the evolution of the income distribution and efficiency depends, not surprisingly, on the shape of the production function. The simplest case is that of constant returns in production. In this case, inequality remains unchanged over time, and production and investment is always efficient. With diminishing returns, greater inequality can lead to less investment and less growth, because the production function is concave. However, inequality falls over time and in the long run no one is credit constrained, although we do not necessarily get full wealth convergence. The long run interest rate converges to its first best level, and hence investment is efficient. To see why this must be the case, note first that because of diminishing returns the poor always have more to gain from borrowing and investing than the rich. In other words, the rich must be lending to the poor. As long as the poor are credit constrained, they will earn higher returns on the marginal dollar than their lenders, i.e., the rich (that is what it means to be credit constrained). As a result, they will accumulate wealth faster than the rich and we will see convergence. This process will only stop when the poor are no longer credit constrained, i.e., they are rich enough to be able to invest as much as they want. With increasing returns, inequality increases over time; we converge to a Gini coefficient of 1. Wealth becomes more and more concentrated with only the richest borrowing and investing. Because there are increasing returns, this is also the first best outcome. The logic of this result is very similar to the previous one: Now it is the rich who will be borrowing and the poor who will be lending, with the implication that the rich are the ones who are credit constrained and the ones earning high marginal returns. Therefore, they will accumulate wealth faster and wealth becomes increasingly concentrated. Finally we consider the case of “S-shaped” production functions, which are production functions that are initially convex and then concave. The Cobb–Douglas with an initial set-up cost discussed at length in Section 5.2 is a special case of this kind of technology. What happens in the long run in this model depends on the initial distribution of income. When the distribution is such that most people in the economy can afford to invest in the concave part of the production function, the economy converges to a situation that is isomorphic to the diminishing returns case, with the entire population “escaping” the convex region of the production function. The more unusual case is the one where some people start too poor to invest in the concave region of the production function. The poorer among such people will earn very low returns if they were to invest and therefore will prefer to be lenders. Now, as
Ch. 7: Growth Theory through the Lens of Development Economics
537
long as the interest rate on savings is less than 1/δ, they will decumulate capital (since the interest is less than the discount factor) and eventually their wealth will go to zero. On the other hand, anyone in this economy who started rich enough to want to borrow will stay rich, even though they are also dissaving, in part because at the same time they benefit from the low interest rates. The economy will converge to a steady state where the interest rate is 1/δ, those who started rich continue to be rich and those who started poor remain poor (in fact have zero wealth). This is classic poverty trap: Moreover, since no one escapes from poverty, nor falls into it, there is a continuum of such poverty traps in this model. This kind of multiplicity is, however, fragile with respect to the introduction of random shocks that allow some of the poor to escape poverty and impoverish some of the rich. Even in a world with such shocks there can be more than one steady state: The reason is that the presence of lots of poor people drives down interest rates, and low interest rates make it harder for the poor to save up to escape poverty even with the help of a positive shock. As a result, in an economy that starts with lots of poor people, a greater fraction of people may remain poor. The key to this multiplicity is the endogeneity of the interest rate. It is the pecuniary externality that the poor inflict on other poor people that sustains it. This is why such poverty traps are sometimes called collective poverty traps, in contrast to the individual poverty traps described above. The investigation of the evolution of income distribution in models with credit constraints and endogenous interest rates goes back to Aghion and Bolton (1997). Matsuyama (2000, 2003) and Piketty (1997) emphasize the potential for collective poverty traps in a variant of this model, without the forward-looking savings decisions. This class of models is a part of a broader group of models which study the simultaneous evolution of the occupational structure, factor prices and the wealth distribution in a model with credit constraints. Loury (1981) studied this class of models and showed that in the long run the neo-classical predictions tend to hold as long as the production function is concave. Dasgupta and Ray (1986) and Galor and Zeira (1993) provide examples of individual poverty traps in the presence of credit constraints and S-shaped production functions. Banerjee and Newman (1993) show the possibility of a collective poverty trap in a model with a S-shaped production function which is driven by the endogeneity of the wage – essentially high wages allow workers to become entrepreneurs easily, which keeps the demand for labor, and hence wages, high. Recent work by Buera (2003) shows that the multiplicity results in Banerjee and Newman survive in an environment where savings is based on expectations of future returns.40 Ghatak, Morelli and Sjostrom (2001, 2002) and Mookherjee and Ray (2002, 2003) explore related but slightly different sources of individual and collective poverty traps.
40 On the possibility of collective poverty traps, see also Lloyd-Ellis and Bernhardt (2000), and Mookherjee
and Ray (2002, 2003).
538
A.V. Banerjee and E. Duflo
6.2. Can we take this model to the data? Models like the one we just developed (as well as political economy models that we do not discuss here41 ) have been invoked as motivation for a large empirical literature on the relationship between inequality and growth in cross-country data. In 1996, Benabou cited 16 studies on the question, and the number has been growing rapidly since then, in part due to the availability of more complete data sets, due to the effort of Deininger and Squire [see Deininger and Squire (1996)], expanded by the World Institute for Development Economics Research (WIDER). However, it is not clear that if we were to take this class of models seriously, they would justify estimating relationships like the ones that are in the literature: First because the exact form of the predicted relationship between inequality and growth depends on the shape of the production function. Imposing the assumption that there are diminishing returns helps in this respect, but with this assumption functional form issues loom large. Finally, it is not clear how, given the model’s structure, we can avoid running into serious identification problems. In this section, we evaluate whether, given these concerns, estimating the relationship between inequality and growth in a cross-country data set remains useful. Having concluded that it has, at best, very limited use, we discuss an alternative approach based on calibrating non-aggregative models using micro data. 6.2.1. What are the empirical implications of the above model? Functional form issues With constant returns to scale, distribution is irrelevant for growth. With diminishing returns, an exogenous mean-preserving spread in the wealth distribution in this economy will reduce future wealth and, by implication, the growth rate. However, the impact depends on the level of wealth in the economy: Once the economy is rich enough that everyone can afford the optimal level of investment, inequality should not matter. The estimated relationship between inequality and growth should therefore allow for an interaction term between inequality and mean income. Moreover, an economy closer to the steady state has both lower inequality and lower growth. This has two implications for the estimation of the inequality growth relationship. First, the fact that the economy becomes more equal as it grows tends to generate a spurious positive relation between growth and inequality, both in the cross-section as well as in time-series. As a result, both the cross-sectional and the first differenced (or fixed effects) estimates of the effect of inequality on growth run the risk of being biased upwards, compared to the true negative relation that we might have found if we had compared economies at the same mean wealth levels. Moreover, consider a variant of the model where there are occasional shocks that increase inequality. Since the natural
41 See Alesina and Rodrik (1994), Persson and Tabellini (1991) and Benhabib and Rustichini (1998). For a
contrary point of view, arguing that the premise of the political economy model argument does not hold true in the data, see Benabou (1996).
Ch. 7: Growth Theory through the Lens of Development Economics
539
tendency of the economy is towards convergence, we should expect to see two types of changes in inequality: Exogenous shocks that increase inequality and therefore reduce growth, and endogenous reductions in inequality that are also associated with a fall in the growth rate. In other words, measured changes in inequality in either direction will be associated with a fall in growth. Controlling properly for the effect of mean wealth (or mean income), is therefore vital for getting meaningful results. The usual procedure is to control linearly (as in most other growth regressions) for the mean income level at the beginning of the period. It is, however, not clear that there is any good reason why the true effect should be linear. Moreover, it seems plausible that different economies will typically have different λs, and therefore will converge at different rates. The model also tells us that while initial distribution matters for the growth rate, it only matters in the short run. Over a long enough period, two economies starting at the same mean wealth level will exhibit the same average growth rate. In other words, the length of the time period over which growth is measured will affect the strength of the relationship between inequality and growth. The preceding discussion assumed that the interest rates converged. As we noted, that does not need to be the case. If we do not assume it, variants of the simple concave economy may no longer converge, even in the weaker sense of the long-run mean wealth being independent of the initial distribution of wealth. Intuitively, poor economies will tend to have high interest rates, and this in turn will make capital accumulation difficult (note that λ < 0) and tend to keep the economy poor.42 This effect reinforces the claim made above that inequality matters most in the poorest economies.43 This economy can have a number of distinct steady states that are each locally isolated. This means that small changes in inequality can cause the economy to move towards a different and further away steady state, making it more likely that the relationship will be non-linear. With increasing returns, growth rates increase with a mean preserving spread in income. As the economy grows, it also becomes more unequal. Interpreting the relationship between inequality and growth is difficult even after controlling for convergence. In the S-shaped returns case, the relationship between inequality and growth can be negative or positive depending on the initial distribution, and the size of the increase. For example, if everybody is very poor (on the left of the convex zone), a small increase in inequality will reduce growth, but increasing inequality enough may push more people to the point where they are able to take advantage of the more efficient technology, and increases in inequality will increase growth. The relation between inequality and
42 See Piketty (1997). For a more general discussion of the issue of convergence in this class of models, see
Banerjee and Newman (1993). 43 There is, however, a counteracting effect: Poorer economies with high levels of inequality may actually
have low interest rates because a few people may own more wealth than they can invest in their own firms, and the rest may be too poor to borrow. For a model where this effect plays an important role, see Aghion and Bolton (1997).
540
A.V. Banerjee and E. Duflo
growth delivered by this model is clearly non-monotonic. Moreover, the strong convergence property does not hold in general. In other words, the growth rate of wealth may jump up once the economy is rich enough, with the obvious implication that economies with higher mean wealth will not necessarily grow more slowly. In other words, the effect of mean wealth, that is the so-called convergence effect, may not be monotonic in this economy. Linearly controlling for mean wealth therefore does not guarantee that we will get the correct estimate of the effect of inequality. It is worth noting that this economy will have a connected continuum of steady states. This means that after a shock the economy will not typically return to the same steady state. However, since it does converge to a nearby steady state, this is not an additional source of non-linearity. Identification issues Even if we could agree on a specification that is worth estimating, it is not clear how we can use cross-country data to estimate it. Countries, like individuals, are different from each other. Even in a world of perfect capital markets, countries can have very different distributions of wealth because, for example, they have different distributions of ability. There is no causal effect of inequality on growth in this case, but they could be correlated for other reasons. For example, cultural structures (such as a caste system) may restrict occupational choices and therefore may not allow individuals to make proper use of their talents, causing both higher inequality and lower growth. Conversely, if countries use technologies that are differently intensive in skilled labor, those countries using the more skill intensive technology can have both more inequality and faster growth. As we discussed in detail above, countries have different kinds of financial institutions, implying differences in the λ’s in our model. Our basic model would predict that the country with the better capital markets is likely both to be more equal and to grow faster (at least once we control for the mean level of income). The correlation between inequality and growth will therefore be a downwards-biased estimate of the causal parameter, if the quality of financial institutions differs across countries.44 If these country specific effects were additive, one could control for them by including a country fixed-effect in the estimated relationship (or by estimating the model in first difference). This strategy will be valid only under the assumption that changes in inequality are unrelated to unobservable country characteristics that are correlated with changes in the growth rate. While this is a convenient assumption, it has no reason to hold in general. For example, skill-biased technological progress will lead both to a change in inequality and a change in growth rates, causing a spurious positive correlation between the two. To make matters worse, we have to recognize the fact that λ itself (and therefore the effect of inequality on growth at a given point in time) may be 44 Allowing λ to vary also implies that the causal effects of inequality will vary with financial development
(which is how Barro (2000) explains his results). The OLS coefficient is therefore a weighted average of different parameters, where the weights are the country-specific contributions to the overall variance in inequality [Krueger and Lindahl (2001)]. It is not at all clear that we are particularly interested in this set of weights.
Ch. 7: Growth Theory through the Lens of Development Economics
541
varying over time as a result of monetary policies or financial development, and may itself be endogenous to the growth process.45 The more general point that comes out of the discussion above is that unless we assume capital markets are extremely efficient (which, in any case, removes one of the important sources of the effect of inequality), changes in inequality will be partly endogenous and related to country characteristics which are themselves related to changes in the growth rate. Identifying the effect of inequality by including a country fixed-effect would not necessarily solve all the endogeneity problems. Moreover, as we discussed above, the theory suggests that the specification should allow for non-linear functional forms, and interaction effects, which will be difficult to accommodate with a fixed effect specification. 6.2.2. Empirical evidence The preceding discussion suggests that empirical exercises using aggregate, crosscountry data to estimate the impact of inequality and growth will be extremely difficult to interpret. The results are also likely to be sensitive to the choice of specification. This may explain the variety of results present in the literature. A long literature [see Benabou (1996) for a survey] estimated a long run equation, with growth between 1990 and 1960 (say) regressed on income in 1960, a set of control variables, and inequality in 1960. Estimating these equations tended to generate negative coefficients for inequality. As the discussion in the previous subsection suggests, there are many reasons to think that this relationship may be biased upward or downwards. To address this problem, Li and Zou (1998) and Forbes (2000) used the Deininger and Squire data set to focus on the impact of inequality on short run (5 years) growth, and introduced a linear fixed effect.46 The results change rather dramatically: The coefficient of inequality in this specification is positive, and significant. Finally, Barro (2000) used the same short frequency data (he is focusing on ten-year intervals), but does not introduce a fixed effect. He finds that inequality is negatively associated with growth in the poorer countries, and positively in rich countries. Banerjee and Duflo (2003) investigate whether there is any reason to worry about the non-linearities that the theory suggests should be present. They find that when growth (or changes in growth) is regressed non-parametrically on changes in inequality, the relationship is an inverted U-shape. There is also a non-linear relationship between past inequality and the magnitudes of changes in inequality. Finally, there seems to be a negative relationship between growth rates and inequality lagged one period. These facts taken together, and in particular the non-linearities in these relationships (rather than the
45 See Acemoglu and Zilibotti (1997), and Greenwood and Jovanovic (1999), for theories of growth with
endogenous financial development. 46 Forbes (2000) also corrects for the bias introduced by introducing a lagged variable in a fixed effect spec-
ification by using the GMM estimator developed by Arellano and Bond (1991).
542
A.V. Banerjee and E. Duflo
variation in samples or control variables), account for the different results obtained by different authors using different specifications. Townsend and Ueda (2003) illustrate very clearly that this diversity of results is likely to come from the functional form and identification problems we just discussed. They simulate the 30 year evolution for 1,000 economies based on a model similar to the ones we describe in this section, with non-linear individual production function and credit constraints. The economies start in 1976, with a distribution of wealth calibrated to match the Thai economy in the same year. They then introduce aggregate and individual level shocks, and run regressions similar to the regressions run in the literature. Using the 1985 year as the “base year”, they replicate the findings of the long run regressions. Using 1980 as the base year, they do not replicate those results. A regression similar to that of Forbes (2000) finds either a positive or negative relationship, depending on sampling decisions. This exercise clearly shows that aggregate cross-country regressions are the wrong tool to evaluate the pertinence of this class of models. 6.3. Where do we go from here? The discussion on functional form and identification, coupled with the empirical evidence of non-linearities even in very simple exercises, suggests that cross-country regressions are unlikely to be able to shed any meaningful light on the empirical relevance of models that integrate credit constraints and other imperfections of the credit markets. This is made worse by the poor quality of the aggregate data, despite the considerable efforts to produce consistent and reliable data sets. This contrasts with the increased availability of large, good quality, micro-economic data sets, which allow for testing specific hypotheses and derive credible identifying restrictions from theory and exogenous sources of variation. Throughout this chapter, we quoted many studies using micro-economic data which tested the micro-foundations for the models we discussed in this section. Even a series of convincing micro-empirical studies will not be enough to give us an overall sense of how, together, they generate aggregate growth, the dynamics of income distribution, and the complex relationships between the two. The lessons of development economics will be lost to growth if they are not brought together in an aggregate context. In other words, it is not enough to use them to loosely motivate cross-sectional growth regression exercises – the discussion in this section is but an example of the misleading conclusions to which this can lead. An alternative that seems likely to be much more fruitful is to try to build macroeconomic models that incorporate the features we discussed, and to use the results from the microeconomic studies as parameters in calibration exercises. The exercise we performed in Section 5 of this chapter is an illustration of the kind of work that we can hope to do. There are a number of recent papers that in some ways go further in this direction than we have gone. In particular, Quadrini (1999) and Cagetti and De Nardi (2003), for the U.S., and Paulson and Townsend (2004), for Thailand, try to calibrate a model with credit constraints to understand the correlation between wealth and the
Ch. 7: Growth Theory through the Lens of Development Economics
543
probability of becoming an entrepreneur. The paper by Buera (2003), mentioned above, emphasizes the fact that the long run correlation between wealth and entrepreneurship is weaker than the short run correlation, because, as noted by Skiba (1978), Deaton (1992), Aiyagari (1994) and Carroll (1997), those who are credit constrained now but want to invest in the future have a very strong incentive to save. This, Buera points out, reduces the ultimate efficiency cost of imperfect credit markets, though in spite of this, the person with the median ability level and the median starting wealth loses about 18% of lifetime welfare because of the credit constraints. Caselli and Gennaioli (2002) offer a slightly different calibration: Like Buera, they are worried about the fact that with credit constraints the biggest firms may not be run by the best entrepreneurs. This can be a source of very large productivity losses in the short run. However, since the best entrepreneurs will make the most money, in the long run their firms would necessarily become the largest, unless they died young. They show that even with this limiting factor, reasonable death rates would imply a 20% loss of productivity when we compare an economy without credit constraints with one that has them. The calibrations so far have not attempted to see if the path of wealth distribution that results from calibrating this type of model matches the data. Our exercise above, for example, tries to match the distribution of firm sizes at a point of time, but says nothing about the path, while Buera does not try to match the data. The one exception is the papers by Robert Townsend and his collaborators based on Thai data [Jeong and Townsend (2003), Townsend and Ueda (2003)]. These papers, as well as those mentioned in the previous paragraphs, start from the assumption that every firm has a single, usually strictly concave, production technology. The only fixed cost comes from the fact that the firm needs an entrepreneur. As we saw above, this model does not do very well in terms of explaining the cross-sectional variation in the firm sector or the overall productivity gap, as compared to a model with a small number of alternative technologies and varying fixed costs. More generally, we need both a better empirical understanding of where the most important sources of inefficiency lie and better integration of this understanding when we assess the predictions of growth theory. And perhaps above all, we need better growth theory: Our exercise at the beginning of this section was intended to advertise the possibility of a growth theory that does not assume aggregation. While we attempted to link the results to some relatively general properties of the production function, our analysis relies heavily on the fact that the inefficiency we assumed was in the credit market and that this took the form of a credit limit that was linear in wealth. One can easily imagine other ways for the credit market to be imperfect and other results from such models. Moreover, while the class of production technologies covered by our model was broader than usual, it does not include the (multiple-fixed-cost) technology that the previous section advocates. There are, of course, other types of non-aggregative models: There are some examples of non-aggregative growth models that build on the inefficiency that comes from
544
A.V. Banerjee and E. Duflo
poorly functioning insurance markets.47 There are also interesting attempts to build growth models that emphasize the fact that some people are favored by the government while others are not, and especially the fact that this changes over time in some predictable way (see Roland Benabou’s contribution to this volume). Some interesting recent work has been done on the dynamic interplay between growth and political institutions (see the chapter by Acemoglu, Johnson and Robinson in this volume) as well as between growth and social institutions [see Oded Galor’s contribution to this volume, as well as Cole, Mailath and Postlewaite (1992, 1998, 2001)]. However, even more than in the case of the literature on credit markets and growth, it is not clear how much the insights from these models rely on specific details of how the environment or the imperfection was modeled and to what extent they can be seen as robust properties of this entire class of models. There are also areas where growth theory has not really reached: We have no models that, for example, incorporate reputation-building or learning into growth theory. The same can be said about the entire class of behavioral models of underinvestment. Finally, there is the open question of whether we gain anything by building grand models that incorporate all these different reasons for inefficiency in a single model. To answer this we would need to assess whether the fact that different forms of inefficiency interact with each other has empirically important consequences. This is an exciting time to think about growth. We are beginning to see the contours of a new vision, both more rooted in evidence and more ambitious in its theorizing.
Acknowledgements The authors are grateful to Pranab Bardhan, Michael Kremer, Rohini Pande, Chris Udry and Ivan Werning for helpful conversations and Philippe Aghion and Seema Jayachandran for detailed comments. A part of this material was presented as the Kuznets Memorial Lecture, 2004, at Yale University. We are grateful for the many comments that we received from the audience.
References Acemoglu, D., Angrist, J. (2001). “How large are human-capital externalities? Evidence from compulsory schooling laws”. In: Bernanke, B., Rogoff, K. (Eds.), NBER Macroeconomics Annual 2000, vol. 15. MIT Press, Cambridge and London, pp. 9–59. Acemoglu, D., Johnson, S., Robinson, J. (2001). “The colonial origins of comparative development: An empirical investigation”. American Economic Review 91 (5), 1369–1401.
47 See Banerjee and Newman (1991) for a theoretical model of non-aggregative growth based on imperfect
insurance markets. Deaton and Paxson (1994) investigate some of empirical implications of this type of model using Taiwanese data. Krussel and Smith (1998) and Angeletos and Calvet (2003) are attempts to calibrate the impact of imperfect insurance on welfare and growth.
Ch. 7: Growth Theory through the Lens of Development Economics
545
Acemoglu, D., Zilibotti, F. (1997). “Was Prometheus unbound by chance? Risk, diversification, and growth”. Journal of Political Economy 105 (4), 709–751. Acemoglu, D., Zilibotti, F. (2001). “Productivity differences”. Quarterly Journal of Economics 116 (2), 563– 606. Aghion, P., Bolton, P. (1997). “A trickle-down theory of growth and development with debt overhang”. Review of Economic Studies 64 (2), 151–172. Aghion, P., Howitt, P. (1992). “A model of growth through creative destruction”. Econometrica 60 (2), 323– 351. Aghion, P., Howitt, P., Mayer-Foulkes, D. (2004). “The effect of financial development on convergence: Theory and evidence”. Working Paper 10358, National Bureau of Economic Research. Aiyagari, S.R. (1994). “Uninsured idiosyncratic risk and aggregate saving”. Quarterly Journal of Economics 109, 569–684. Aleem, I. (1990). “Imperfect information, screening and the costs of informal lending: A study of a rural credit market in Pakistan”. World Bank Economic Review 3, 329–349. Alesina, A., Rodrik, D. (1994). “Distributive politics and economic growth”. Quarterly Journal of Economics 109 (2), 465–490. Anderson, S., Baland, J.-M. (2002). “The economics of ROSCAs and intrahousehold resource allocation”. Quarterly Journal of Economics 117 (3), 963–995. Angeletos, G.-M., Calvet, L. (2003). “Idiosyncratic production risk, growth and the business cycle”. Mimeo, MIT. Aportela, F. (1998). “The effects of financial access on savings by low-income people”. Mimeo, MIT. Arellano, M., Bond, S. (1991). “Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations”. Review of Economic Studies 58 (2), 277–297. Ashraf, N., Karlan, D.S., Yin, W. (2004). “Tying Odysseus to the mast: Evidence from a commitment savings product in the Philippines”. Mimeo, Harvard University. Banerjee, A.V. (1992). “A simple model of herd behavior”. Quarterly Journal of Economics 117 (3), 797–817. Banerjee, A.V. (2003). “Contracting constraints, credit markets and economic development”. In: Hansen, L., Dewatripont, M., Turnovsky, S. (Eds.), Advances in Economics and Econometrics: Theory and Applications, Eighth World Congress, vol. III. Cambridge University Press. Banerjee, A.V. (2004). “Educational policy and the economics of the family”. Journal of Development Economics 74 (1), 3–32. Banerjee, A.V. (2005). “The two poverties”. In: Dercan, S. (Ed.), Insurance Against Poverty. Oxford University Press, pp. 59–75. Banerjee, A.V., Duflo, E. (2000). “Reputation effects and the limits of contracting: A study of the Indian software industry”. Quarterly Journal of Economics 115 (3), 989–1017. Banerjee, A.V., Duflo, E. (2003). “Inequality and growth: What can the data say?”. Journal of Economic Growth 8, 267–299. Banerjee, A.V., Duflo, E. (2004). “Do firms want to borrow more? Testing credit constraints using a directed lending program”. Mimeo, MIT. Also BREAD WP 2003-5, 2003. Banerjee, A.V., Duflo, E., Munshi, K. (2003). “The (mis)allocation of capital”. Journal of the European Economic Association 1 (2–3), 484–494. Banerjee, A.V., Gertler, P., Ghatak, M. (2002). “Empowerment and efficiency: Tenancy reform in West Bengal”. Journal of Political Economy 110 (2), 239–280. Banerjee, A.V., Munshi, K. (2004). “How efficiently is capital allocated? Evidence from the knitted garment industry in Tirupur”. Review of Economic Studies 71 (1), 19–42. Banerjee, A.V., Newman, A. (1991). “Risk bearing and the theory of income distribution”. Review of Economic Studies 58 (2), 211–235. Banerjee, A.V., Newman, A. (1993). “Occupational choice and the process of development”. Journal of Political Economy 101 (2), 274–298. Bardhan, P. (1969). “Equilibrium growth in a model with economic obsolescence of machines”. Quarterly Journal of Economics 83 (2), 312–323.
546
A.V. Banerjee and E. Duflo
Bardhan, P., Udry, C. (1999). Development Microeconomics. Oxford University Press, Oxford, New York. Barro, R.J. (1974). “Are government bonds net wealth?”. Journal of Political Economy 82 (6), 1095–1117. Barro, R.J. (1991). “Economic growth in a cross section of countries”. Quarterly Journal of Economics 106 (2), 407–443. Barro, R.J. (2000). “Inequality and growth in a panel of countries”. Journal of Economic Growth 5 (1), 5–32. Barro, R.J., Lee, J.-W. (2000). “International data on educational attainment updates and implications”. Working Paper 7911, National Bureau of Economic Research, September. Barro, R.J., Sala-i-Martin, X. (1995). Economic Growth. McGraw-Hill, New York. Basta, S.S., Soekirman, D.S., Karyadi, D., Scrimshaw, N.S. (1979). “Iron deficiency anemia and the productivity of adult males in Indonesia”. American Journal of Clinical Nutrition 32 (4), 916–925. Becker, G. (1981). A Treatise on the Family. Harvard University Press, Cambridge, MA. Benabou, R. (1996). “Inequality and growth”. In: Bernanke, B., Rotemberg, J.J. (Eds.), NBER Macroeconomics Annual 1996. MIT Press, Cambridge and London, pp. 11–73. Benhabib, J., Rustichini, A. (1998). “Social conflict and growth”. Journal of Economic Growth 1 (1), 143– 158. Benhabib, J., Spiegel, M.M. (1994). “The role of human capital in economic development: Evidence from aggregate cross-country data”. Journal of Monetary Economics 34 (2), 143–174. Bennell, P. (1996). “Rates of return to education: Does the conventional pattern prevail in sub-Saharan Africa?”. World Development 24 (1), 183–199. Bertrand, M., Karlan, D., Mullainathan, S., Zinman, J. (2004). “Pricing psychology.” Mimeo, Harvard University. Besley, T. (1995). “Savings, credit and insurance”. In: Behrman, J., Srinivasan, T.N. (Eds.), Handbook of Development Economics, vol. 3A. Elsevier Science, Amsterdam, pp. 2123–2207. Chapter 6. Besley, T., Case, A. (1994). “Diffusion as a learning process: Evidence from HYV cotton”. Discussion Paper 174, RPDS, Princeton University. Bigsten, A., Isaksson, A., Soderbom, M., Collier, P. (2000). “Rates of return on physical and human capital in Africa’s manufacturing sector”. Economic Development and Cultural Change 48 (4), 801–827. Bikhchandani, S., Hirshleifer, D., Welch, I. (1992). “A theory of fads, fashion, custom, and cultural change as informational cascades”. Journal of Political Economy 100 (5), 992–1026. Bils, M., Klenow, P. (2000). “Does schooling cause growth?”. American Economic Review 90 (5), 1160– 1183. Binswanger, H., Rosenzweig, M. (1986). “Behavioural and material determinants of production relations in agriculture”. Journal of Development Studies 22 (3), 503–539. Blanchflower, D., Oswald, A. (1998). “What makes an entrepreneur?”. Journal of Labor Economics 16 (1), 26–60. Bottomley, A. (1963). “The cost of administering private loans in underdeveloped rural areas”. Oxford Economic Papers 15 (2), 154–163. Bourguignon, F., Chiappori, P.-A. (1992). “Collective models of household behavior”. European Economic Review 36, 355–364. Buera, F. (2003). “A dynamic model of entrepreneurship with borrowing constraints”. Mimeo, University of Chicago. Cagetti, M., De Nardi, M. (2003). “Entrepreneurship, frictions and wealth”. Staff Report 324, Federal Reserve Bank of Minneapolis. Carroll, C. (1997). “Buffer-stock saving and the life cycle/permanent income hypothesis”. Quarterly Journal of Economics 112, 1–56. Carvalho, I. (2000). “Household income as a determinant of child labor and school enrollment in Brazil: Evidence from a social security reform”. Mimeo, MIT. Case, A., Lin, I.-F., McLanahan, S. (2000). “How hungry is the selfish gene?”. Economic Journal 110 (466), 781–804. Case, A., Paxson, C. (2001). “Mothers and others: Who invests in children’s health?”. Journal of Health Economics 20 (3), 301–328.
Ch. 7: Growth Theory through the Lens of Development Economics
547
Case, A., Paxson, C., Ableidinger, J. (2002). “Orphans in Africa”. Working Paper 9213, National Bureau of Economic Research. Caselli, F. (2005). “Accounting for cross-country income differences”. In: Aghion, P., Durlauf, S. (Eds.), Handbook of Economic Growth. Elsevier, Amsterdam. Caselli, F., Gennaioli, N. (2002). “Dynastic management”. Mimeo, Harvard University. Caselli, F., Gennaioli, N. (2004). “Dynastic management, legal reform and regulatory reform”. Mimeo, London School of Economics. Coate, S., Ravallion, M. (1993). “Reciprocity without commitment: Characterization and performance of informal insurance arrangements”. Journal of Development Economics 40, 1–24. Cole, H., Mailath, G., Postlewaite, A. (1992). “Social norms, savings behavior and growth”. Journal of Political Economy 100 (6), 1092–1126. Cole, H., Mailath, G., Postlewaite, A. (1998). “Class systems and the enforcement of social norms”. Journal of Public Economics 70 (1), 5–35. Cole, H., Mailath, G., Postlewaite, A. (2001). “Investment and concern for relative position”. Review of Economic Design 6, 241–261. Commission on Macroeconomics, and Health (2001). Macroeconomics and Health: Investing in Health for Economic Development: Report. World Health Organization, Geneva. Conley, T., Udry, C. (2003). “Learning about a new technology: Pineapple in Ghana”. Mimeo, Northwestern University. Dasgupta, A. (1989). Reports on Informal Credit Markets in India: Summary. National Institute of Public Finance and Policy, New Delhi. Dasgupta, P., Ray, D. (1986). “Inequality as a determinant of malnutrition and unemployment: Theory”. The Economic Journal 96 (384), 1011–1034. Deaton, A. (1992). Understanding Consumption. Oxford University Press, Oxford. Deaton, A. (1997). The Analysis of Household Surveys. World Bank, International Bank for Reconstruction and Development. Deaton, A., Paxson, C. (1994). “Intertemporal choice and inequality”. Journal of Political Economy 1–2 (3), 437–467. Deininger, K., Squire, L. (1996). “A new data set measuring income inequality”. World Bank Economic Review 10, 565–591. Djankov, S., La Porta, R., Lopez de Silanes, F., Shleifer, A. (2002). “The regulation of entry”. Quarterly Journal of Economics 117 (1), 1–37. Djankov, S., La Porta, R., Lopez de Silanes, F., Shleifer, A. (2003). “Courts”. Quarterly Journal of Economics 118 (2), 453–518. Do, T., Iyer, L. (2003). “Land rights and economic development: Evidence from Vietnam”. Mimeo, Harvard Business School. Dreze, J., Kingdon, G.G. (2001). “School participation in rural India”. Review of Development Economics 5 (1), 1–24. Duflo, E. (2004). “The medium run effects of educational expansion: Evidence from a large school construction program in Indonesia”. Journal of Development Economics 74 (1), 163–197. Duflo, E. (2003). “Poor but rational”. Mimeo, MIT. Duflo, E., Kremer, M., Robinson, J. (2003). “Understanding technology adoption: Fertilizer in Western Kenya, preliminary results from field experiments”. Mimeo, MIT. Durlauf, S. (1993). “Nonergodic economic growth”. Review of Economic Studies 60 (2), 349–366. Eaton, J., Kortum, S. (2001). “Trade in capital goods”. European Economic Review 45 (7), 1195–1235. Edmonds, E. (2004). “Does illiquidity alter child labor and schooling decisions? Evidence from household responses to anticipated cash transfers in South Africa”. Working Paper 10265, National Bureau of Economic Research. Ellison, G., Fudenberg, D. (1993). “Rules of thumbs for social learning”. Journal of Political Economy 101 (4), 93–126. Ellison, G., Glaeser, E. (1997). “Geographic concentration in U.S. manufacturing industries: A dartboard approach”. Journal of Political Economy 105 (5), 889–927.
548
A.V. Banerjee and E. Duflo
Ellison, G., Glaeser, E. (1999). “Geographic concentration of industry: Does natural advantage explain agglomeration?”. American Economic Review 89 (2), 311–316. Eswaran, M., Kotwal, A. (1985). “A theory of contractual structure in agriculture”. American Economic Review 75 (3), 352–367. Fafchamps, M. (2000). “Ethnicity and credit in African manufacturing”. Journal of Development Economics 61, 205–235. Fafchamps, M., Lund, S. (2003). “Risk-sharing networks in rural Philippines”. Review of Economic Studies 71 (2), 261–287. Fazzari, S., Hubbard, G., Petersen, B. (1988). “Financing constraints and corporate investment”. Brookings Papers on Economic Activity 0 (1), 141–195. Forbes, K.J. (2000). “A reassessment of the relationship between inequality and growth”. American Economic Review 90 (4), 869–887. Foster, A.D., Rosenzweig, M.R. (1995). “Learning by doing and learning from others: Human capital and technical change in agriculture”. Journal of Political Economy 103 (6), 1176–1209. Foster, A.D., Rosenzweig, M.R. (1996). “Technical change and human-capital returns and investments: Evidence from the green revolution”. American Economic Review 86 (4), 931–953. Foster, A.D., Rosenzweig, M.R. (1999). “Missing women, the marriage market and economic growth”. Mimeo, University of Pennsylvania. Foster, A.D., Rosenzweig, M. (2000). “Technological change and the distribution of schooling: Evidence from green revolution in India”. Mimeo, Brown University. Freeman, R. (1986). “Demand for education”. In: Ashenfelter, O., Layard, A. (Eds.), Handbook of Labor Economics, vol. 1. Elsevier, Amsterdam. Chapter 6. Freeman, R., Oostendorp, R. (2001). “The occupational wages around the world data file”. International Labour Review 140 (4), 379–401. Galor, O., Zeira, J. (1993). “Income distribution and macroeconomics”. Review of Economic Studies 60 (1), 35–52. Gelos, R.G., Werner, A. (2002). “Financial liberalization, credit constraints, and collateral: Investment in the Mexican manufacturing sector”. Journal of Development Economics 67 (1), 1–27. Gertler, P., Gruber, J. (2002). “Insuring consumption against illness”. American Economic Review 92 (1), 51–76. Gertler, P.J., Boyce, S. (2002). “An experiment in incentive-based welfare: The impact of PROGESA on health in Mexico”. Mimeo, University of California, Berkeley. Ghatak, M., Morelli, M., Sjostrom, T. (2001). “Occupational choice and dynamic incentives”. Review of Economic Studies 68 (4), 781–810. Ghatak, M., Morelli, M., Sjostrom, T. (2002). “Credit rationing, wealth inequality, and allocation of talent”. Mimeo, London School of Economics. Ghate, P. (1992). Informal Finance: Some Findings from Asia. Oxford University Press for the Asian Development Bank, Oxford. Goldstein, M., Udry, C. (1999). “Agricultural innovation and resource management in Ghana”. Mimeo, Yale University. Final Report to IFPRI under MP17. Goldstein, M., Udry, C. (2002). “Gender, land rights and agriculture in Ghana”. Mimeo, Yale University. Greenwood, J., Jovanovic, B. (1999). “The information technology revolution and the stock market”. American Economic Review 89 (2), 116–122. Grossman, G., Helpman, E. (1991). Innovation and Growth in the Global Economy. MIT Press, Cambridge, MA. Gugerty, M.K. (2000). “You can’t save alone: Testing theories of rotating savings and credit associations”. Mimeo, Harvard University. Hart, O., Moore, J. (1994). “A theory of debt based on inalienability of human capital”. Quarterly Journal of Economics 109, 841–879. Heston, A., Summers, R., Aten, B. (2002). “Penn World Table version 6.1”. Technical Report, Center for International Comparisons at the University of Pennsylvania.
Ch. 7: Growth Theory through the Lens of Development Economics
549
Hoff, K., Stiglitz, J. (1998). “Moneylenders and bankers: Price-increasing subsidies in a monopolistically competitive market”. Journal of Development Economics 55 (2), 485–518. Howitt, P., Mayer, D. (2002). “R&D, implementation and stagnation: A Schumpeterian theory of convergence clubs.” Working Paper 9104, National Bureau of Economic Research. Hsieh, C.-T. (1999). “Productivity growth and factor prices in East Asia”. American Economic Review 89 (2), 133–138. Hsieh, C.-T., Klenow, P.J. (2003). “Relative prices and relative prosperity”. Mimeo, University of California, Berkeley. Jacoby, H. (2002). “Is there an intrahousehold flypaper effect? Evidence from a school feeding program”. Economic Journal 112 (476), 196–221. Jeong H., Townsend R. (2003). “Growth and inequality: Model evaluation based on an estimation-calibration strategy”. Mimeo, University of Southern California. Johnson, S., McMillan, J., Woodruff, C. (2002). “Property rights and finance”. American Economic Review 92 (5), 1335–1356. Jovanovic, B., Rob, R. (1997). “Solow vs. Solow: Machine prices and development”. Working Paper 5871, National Bureau of Economic Research. Jovanovic, B., Rousseau, P. (2003). “Specific capital and the division of rents”. Mimeo, New York University. Kanbur, R. (1979). “Of risk taking and the personal distribution of income”. Journal of Political Economy 87, 769–797. Karlan, D. (2003). “Using experimental economics to measure social capital and predict financial decisions”. Mimeo, Princeton University. Kihlstrom, R., Laffont, J. (1979). “A general equilibrium entrepreneurial theory of firm formation based on risk aversion”. Journal of Political Economy 87, 719–748. Klenow, P.J., Rodriguez-Clare, A. (1997). “The neoclassical revival in growth economics: Has it gone too far?”. In: Bernanke, B., Rotemberg, J.J. (Eds.), NBER Macroeconomics Annual 1997. MIT Press. Knack, S., Keefer, P. (1995). “Institutions and economic performance: Cross-country tests using alternative institutional measures”. Economics and Politics 7 (3), 207–227. Kremer, M., Miguel, E. (2003). “The illusion of sustainability”. Mimeo, University of California, Berkeley. Kremer, M., Moulin, S., Namunyu, R. (2003). “Decentralization: A cautionary tale”. Mimeo, Harvard University. Krueger, A. (1967). “Factor endowments and per capital income differences among countries”. Economic Journal 78, 641–659. Krueger, A., Lindahl, M. (2001). “Education for growth: Why and for whom?”. Journal of Economic Literature 39 (4), 1101–1136. Krussel, P., Smith, A. (1998). “Income and wealth heterogeneity in the macroeconomy”. Journal of Political Economy 106 (5), 867–896. La Porta, R., Lopez de Silanes, F., Shleifer, A., Vishny, R. (1998). “Law and finance”. Journal of Political Economy 106 (6), 1113–1155. Laffont, J.-J., Matoussi, M.S. (1995). “Moral hazard, financial constraints and sharecropping in El Oulja”. Review of Economic Studies 62 (3), 381–399. Laibson, D. (1991). “Golden eggs and hyperbolic discounting”. Quarterly Journal of Economics 62, 443–477. Levine, R., Renelt, D. (1992). “A sensitivity analysis of cross-country growth regressions”. American Economic Review 82 (4), 942–963. Li, H., Zou, H.-f. (1998). “Income inequality is not harmful for growth: Theory and evidence”. Review of Development Economics 2 (3), 318–334. Li, R., Chen, X., Yan, H., Deurenberg, P., Garby, L., Hautvast, J.G. (1994). “Functional consequences of iron supplementation in iron-deficient female cotton workers in Beijing China”. American Journal of Clinical Nutrition 59, 908–913. Lloyd-Ellis, H., Bernhardt, D. (2000). “Enterprise, inequality and economic development”. Review of Economic Studies 67 (1), 147–169. Long, S.K. (1991). “Do the school nutrition programs supplement household food expenditures?”. Journal of Human Resources 26, 654–678.
550
A.V. Banerjee and E. Duflo
Loury, G.C. (1981). “Intergenerational transfers and the distribution of earnings”. Econometrica 49 (4), 843– 867. Lucas, R. (1990). “Why doesn’t capital flow from rich to poor countries?”. American Economic Review 80 (2), 92–96. Lundberg, S., Pollak, R. (1994). “Noncooperative bargaining models of marriage”. American Economic Review 84 (2), 132–137. Lundberg, S., Pollak, R. (1996). “Bargaining and distribution in marriage”. Journal of Economic Perspectives 10 (4), 139–158. Mankiw, N.G., Romer, D., Weil, D.N. (1992). “A contribution to the empirics of economic growth”. Quarterly Journal of Economics 107 (2), 407–437. Manski, C. (1993). “Identification of exogenous social effects: The reflection problem”. Review of Economic Studies 60, 531–542. Matsuyama, K. (2000). “Endogenous inequality”. Review of Economic Studies 67 (4), 743–759. Matsuyama, K. (2003). “On the rise and fall of class societies”. Mimeo, Northwestern University. Mauro, P. (1995). “Corruption and growth”. Quarterly Journal of Economics 110 (3), 681–712. McKenzie, D., Woodruff, C. (2003). “Do entry costs provide an empirical basis for poverty traps? Evidence from Mexican microenterprises”. Working Paper 2003-20, Bureau for Research in Economic Analysis of Development. McKinsey Global Institute (2001). “India: The growth imperative”. Report, McKinsey Global Institute. Miguel, E., Kremer, M. (2003). “Networks, social learning, and technology adoption: The case of deworming drugs in Kenya”. Working Paper 61. Miguel, E., Kremer, M. (2004). “Worms: Identifying impacts on education and health in the presence of treatment externalities”. Econometrica 72 (1), 159–218. Mookherjee, D., Ray, D. (2002). “Contractual structure and wealth accumulation”. American Economic Review 92 (4), 818–849. Mookherjee, D., Ray, D. (2003). “Persistent inequality”. Review of Economic Studies 70 (2), 369–393. Morduch, J. (1993). “Risk production and saving: Theory and evidence from Indian households”. Mimeo, Harvard University. Munshi, K. (2004). “Social learning in a heterogeneous population: Technology diffusion in the Indian green revolution”. Journal of Development Economics 73 (1), 185–213. Munshi, K., Myaux, J. (2002). “Development as a process of social change: An application to the fertility transition”. Mimeo, Brown University. Munshi, K., Rosenzweig, M. (2004). “Traditional institutions meet the modern world: Caste, gender, and schooling choice in a globalizing economy”. Mimeo, Brown University. Murphy, K., Shleifer, A., Vishny, R. (1989). “Industrialization and the big push”. Journal of Political Economy 97 (5), 1003–1026. Murphy, K., Shleifer, A., Vishny, R. (1995). “The allocation of talent: Implications for growth”. In: Tollison, R., Congleton, R. (Eds.), The Economic Analysis of Rent Seeking. In: Elgar Reference Collection: International Library of Critical Writings in Economics, vol. 49. Ashgate, Elgar, pp. 301–328. Murshid, K. (1992). “Informal credit markets in Bangladesh agriculture: Bane or boon?”. In: Sustainable Agricultural Development: The Role of International Cooperation: Proceedings of the 21st International Conference of Agricultural Economists, Tokyo, Japan, 22–29 August 1991. Dartmouth/Ashgate, Aldershot, UK/Brookfield, VT, pp. 657–668. Newman, A. (1995). “Risk-bearing and ‘Knightian’ entrepreneurship”. Mimeo, Columbia University. Olley, G.S., Pakes, A. (1996). “The dynamics of productivity in the telecommunications equipment industry”. Econometrica 64 (6), 1263–1297. Parente, S., Prescott, E. (1994). “Barriers to technology adoption and development”. Journal of Political Economy 102 (2), 298–321. Parente, S., Prescott, E. (2000). Barriers to Riches: Walras–Pareto Lectures, vol. 3. MIT Press, Cambridge, MA. Paulson, A., Townsend, R. (2004). “Entrepreneurship financial constraints in Thailand”. Journal of Corporate Finance 10 (2), 229–262.
Ch. 7: Growth Theory through the Lens of Development Economics
551
Persson, T., Tabellini, G. (1991). “Is inequality harmful for growth? Theory and evidence”. American Economic Review 48, 600–621. Pessoa, S., Cavalcanti-Ferreira, P., Velloso, F. (2004). “The evolution of international output differences (1960–2000): From factors to productivity”. Mimeo, EPGE-FGV Rio. Piketty, T. (1997). “The dynamics of the wealth distribution and the interest rate with credit rationing”. The Review of Economic Studies 64 (2), 173–189. Powell, C., Grantham-McGregor, S., Elston, M. (1983). “An evaluation of giving the Jamaican government school meal to a class of children”. Human Nutrition: Clinical Nutrition 37C, 381–388. Powell, C., Walker, S., Chang, S., Grantham-McGregor, S. (1998). “Nutrition and education: A randomized trial of the effects of breakfast in rural primary school children”. American Journal of Clinical Nutrition 68, 873–879. Psacharopoulos, G. (1973). Returns to Education: An International Comparison. Jossy Bass–Elsevier, San Francisco. Psacharopoulos, G. (1985). “Returns to education: A further international update and implications”. Journal of Human Resources 20 (4), 583–604. Psacharopoulos, G. (1994). “Returns to investments in education: A global update”. World Development 22 (9), 1325–1343. Psacharopoulos, G., Patrinos, H.A. (2002). “Returns to investment in education: A further update”. Policy Research Working Paper 2881, The World Bank. Qian, N. (2003). “Missing women and the price of tea in China”. Mimeo, MIT. Quadrini, V. (1999). “The importance of entrepreneurship for wealth concentration and mobility”. Review of Income and Wealth 45, 1–19. Rauch, J. (1993). “Productivity gains from geographic concentration of human capital: Evidence from the cities”. Journal of Urban Economics 34 (3), 380–400. Ravallion, M., Chaudhuri, S. (1997). “Risk and insurance in village India: Comment”. Econometrica 65 (1), 171–184. Ray, D. (1998). Development Economics. Princeton University Press, Princeton, NJ. Ray, D. (2003). “Aspirations, poverty and economic change”. Mimeo, New York University. Restuccia, D., Urrutia, C. (2001). “Relative prices and investment rates”. Journal of Monetary Economics 47 (1), 93–121. Rosenstein-Rodan, P.N. (1943). “Problems of industrialization of Eastern and South-Eastern Europe”. Economic Journal 53, 202–211. Rosenzweig, M.R., Binswanger, H. (1993). “Wealth, weather risk and the composition and profitability of agricultural investments”. Economic Journal 103 (416), 56–78. Rosenzweig, M.R., Schultz, T.P. (1982). “Market opportunities, genetic endowments, and intrafamily resource distribution: Child survival in rural India”. American Economic Review 72 (4), 803–815. Rosenzweig, M.R., Wolpin, K.I. (1993). “Credit market constraints, consumption smoothing, and the accumulation of durable production assets in low-income countries: Investments in bullocks in India”. Journal of Political Economy 101 (21), 223–244. Sala-i-Martin, X. (1997). “I just ran four million regressions”. Working Paper 6252, National Bureau of Economic Research. Schultz, T.P. (2004). “School subsidies for the poor: Evaluating the Mexican PROGRESA poverty program”. Journal of Development Economics 74 (1), 199–250. Shaban, R. (1987). “Testing between competing models of sharecropping”. Journal of Political Economy 95 (5), 893–920. Skiba, A.K. (1978). “Optimal growth with a convex-concave production function”. Econometrica 46, 527– 539. Stiglitz, J. (1969). “The effects of income, wealth, and capital gains taxation on risk-taking”. Quarterly Journal of Economics 83 (2), 263–283. Stiglitz, J. (1974). “Incentives and risk sharing in sharecropping”. Review of Economic Studies 41 (2), 219– 255.
552
A.V. Banerjee and E. Duflo
Strauss, J., Thomas, D. (1995). “Human resources: Empirical modeling of household and family decisions”. In: Behrman, J., Srinivasan, T.N. (Eds.), Handbook of Development Economics, vol. 3A. North-Holland, Amsterdam, pp. 1885–2023. Chapter 34. Strauss, J., Thomas, D. (1998). “Health, nutrition, and economic development”. Journal of Economic Literature 36 (2), 766–817. Svensson, J. (1998). “Investment, property rights and political instability: Theory and evidence”. European Economic Review 42 (7), 1317–1341. Thomas, D. (2001). “Health, nutrition and economic prosperity: A microeconomic perspective”. Working Paper, Bulletin of the World Working Paper 7, Working Group 1, WHO Commission on Macroeconomics and Health. Thomas, D., Frankenberg, E. (2002). “Health, nutrition and prosperity: A microeconomic perspective”. Bulletin of the World Health Organization 80 (2), 106–113. Thomas, D., Frankenberg, E., Friedman, J., Habicht, J.-P. (2003). “Iron deficiency and the well being of older adults: Early results from a randomized nutrition intervention”. Mimeo, UCLA. Timberg, T., Aiyar, C.V. (1984). “Informal credit markets in India”. Economic Development and Cultural Change 33 (1), 43–59. Tirole, J. (1996). “A theory of collective reputations (with applications to the persistence of corruption and to firm quality)”. Review of Economic Studies 63 (1), 1–22. Townsend, R. (1994). “Risk and insurance in village India”. Econometrica 62 (4), 539–591. Townsend, R. (1995). “Financial systems in Northern Thai villages”. Quarterly Journal of Economics 110 (4), 1011–1046. Townsend, R., Ueda, K. (2003). “Financial deepening, inequality, and growth: A model-based quantitative evaluation”. Mimeo, University of Chicago. Udry, C. (1990). “Credit markets in Northern Nigeria: Credit as insurance in a rural economy”. World Bank Economic Review 4 (3), 251–269. Udry, C. (1996). “Gender, agricultural production, and the theory of the household”. Journal of Political Economy 101 (5), 1010–1045. Udry, C. (2003). “A note on the returns to capital in a developing country”. Mimeo, Yale University. United Nations Development Program (2001). Human Development Report 2001: Making New Technologies Work for Human Development. Oxford University Press, Oxford and New York. Vermeersch, C. (2002). “School meals, educational achievement and school competition: Evidence from a randomized evaluation”. Mimeo, Harvard University. Young, A. (1995). “The tyranny of numbers: Confronting the statistical realities of the East Asian growth experience”. Quarterly Journal of Economics 110 (3), 641–680. Zeldes, S. (1989). “Consumption and liquidity constraints: An empirical investigation”. Journal of Political Economy 97 (2), 305–346.
Chapter 8
GROWTH ECONOMETRICS STEVEN N. DURLAUF Department of Economics, University of Wisconsin, 1180 Observatory Drive, Madison, WI 53706-1393, USA PAUL A. JOHNSON Department of Economics, Vassar College, 124 Raymond Avenue, Poughkeepsie, NY 12064-0708, USA JONATHAN R.W. TEMPLE Department of Economics, University of Bristol, 8 Woodland Road, Bristol BS8 1TN, UK
Contents Abstract Keywords 1. Introduction 2. Stylized facts 2.1. A long-run view 2.2. Data after 1960 2.3. Differences in levels of GDP per worker 2.4. Growth miracles and disasters 2.5. Convergence? 2.6. The growth slowdown 2.7. Does past growth predict future growth? 2.8. Growth differences by development level and geographic region 2.9. Stagnation and output volatility 2.10. A summary of the stylized facts
3. Cross-country growth regressions: from theory to empirics 3.1. Growth dynamics: basic ideas 3.2. Cross-country growth regressions 3.3. Interpreting errors in growth regressions
4. The convergence hypothesis 4.1. Convergence and initial conditions 4.2. β-convergence 4.2.1. Robustness with respect to choice of control variables
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01008-7
556 557 558 561 562 562 563 565 567 567 568 571 573 575 576 576 578 581 582 582 585 587
556
S.N. Durlauf et al. 4.2.2. Identification and nonlinearity: β-convergence and economic divergence 4.2.3. Endogeneity 4.2.4. Measurement error 4.2.5. Effects of linear approximation 4.3. Distributional approaches to convergence 4.3.1. σ -convergence 4.3.2. Evolution of the world income distribution 4.3.3. Distribution dynamics 4.3.4. Relationship between distributional convergence and the persistence of initial conditions 4.4. Time series approaches to convergence 4.5. Sources of convergence or divergence
5. Statistical models of the growth process 5.1. Specifying explanatory variables in growth regressions 5.2. Parameter heterogeneity 5.3. Nonlinearity and multiple regimes
6. Econometric issues I: Alternative data structures 6.1. 6.2. 6.3. 6.4.
Time series approaches Panel data Event study approaches Endogeneity and instrumental variables
7. Econometric issues II: Data and error properties 7.1. 7.2. 7.3. 7.4. 7.5.
Outliers Measurement error Missing data Heteroskedasticity Cross-section error correlation
8. Conclusions: The future of growth econometrics Acknowledgements Appendix A: Data Key to the 102 countries Extrapolation
Appendix B: Variables in cross-country growth regressions Appendix C: Instrumental variables for Solow growth determinants Appendix D: Instrumental variables for non-Solow growth determinants References
588 589 590 591 592 592 593 596 598 599 604 607 608 616 619 624 624 627 636 637 640 641 641 642 643 643 645 651 651 651 652 652 660 661 663
Abstract This paper provides a survey and synthesis of econometric tools that have been employed to study economic growth. While these tools range across a variety of statistical methods, they are united in the common goals of first, identifying interesting contemporaneous patterns in growth data and second, drawing inferences on long-run economic
Ch. 8: Growth Econometrics
557
outcomes from cross-section and temporal variation in growth. We describe the main stylized facts that have motivated the development of growth econometrics, the major statistical tools that have been employed to provide structural explanations for these facts, and the primary statistical issues that arise in the study of growth data. An important aspect of the survey is attention to the limits that exist in drawing conclusions from growth data, limits that reflect model uncertainty and the general weakness of available data relative to the sorts of questions for which they are employed.
Keywords identification, estimation, parameter heterogeneity, model uncertainty, nonlinearities, convergence, growth determinants JEL classification: C2, C3, O1, O2, O3
558
S.N. Durlauf et al.
The totality of our so-called knowledge or beliefs, from the most causal matters of geography and history to the profoundest laws of atomic physics . . . is a manmade fabric which impinges on experience only along the edges . . . total science is like a field of force whose boundary conditions are experience . . . A conflict with experience on the periphery occasions readjustments in the interior of the field. Reevaluation of some statements entails reevaluation of others, because of their logical interconnections . . . But the total field is so underdetermined by its boundary conditions, experience, that there is much latitude of choice as to what statements to reevaluate in the light of any single contrary experience. W.V.O. Quine1
1. Introduction The empirical study of economic growth occupies a position that is notably uneasy. Understanding the wealth of nations is one of the oldest and most important research agendas in the entire discipline. At the same time, it is also one of the areas in which genuine progress seems hardest to achieve. The contributions of individual papers can often appear slender. Even when the study of growth is viewed in terms of a collective endeavor, the various papers cannot easily be distilled into a consensus that would meet standards of evidence routinely applied in other fields of economics. A traditional defense of empirical growth research would be in terms of expected payoffs. Each time an empirical growth paper is written, the probability of gaining genuine understanding may be low, but the payoff to that understanding is potentially vast. But even this argument relies on being able to discriminate between the status of different pieces of evidence – the good, the bad and the ugly – and this process of discrimination carries many difficulties of its own. Rodriguez and Rodrik (2001) begin their skeptical critique of evidence on trade policy and growth with an apt quote from Mark Twain: “It isn’t what we don’t know that kills us. It’s what we know that ain’t so”. This point applies with especial force in the identification of empirically salient growth determinants. As illustrated in Appendix B of this chapter, approximately as many growth determinants have been proposed as there are countries for which data are available. It is hard to believe that all these determinants are central, yet the embarrassment of riches also makes it hard to identify the subset that truly matters. There are other respects in which it is difficult to reconcile alternative empirical studies, including the functional form posited for the growth process. An important distinction between the neoclassical growth model of Solow (1956) and Swan (1956) and many of the models that have been produced in the endogenous growth theory literature launched by Romer (1986) and Lucas (1988) is that the latter can require the specification of a nonlinear data generating process. But researchers have not yet agreed on the 1 “Two Dogmas of Empiricism”, Philosophical Review, 1951.
Ch. 8: Growth Econometrics
559
empirical specification of growth nonlinearities, or the methods that should be used to distinguish neoclassical and endogenous growth models empirically. These and other difficulties inherent in the empirical study of growth have prompted the field to evolve continuously, and to adopt a wide range of methods. We argue that a sufficiently rich set of statistical tools for the study of growth have been developed and applied that they collectively define an area of growth econometrics. This chapter is designed to provide an overview of the current state of this field. The chapter will both survey the body of econometric and statistical methods that have been brought to bear on growth questions and provide some assessments of the value of these tools. Much of growth econometrics reflects the specialized questions that naturally arise in growth contexts. For example, statistical tools are often used to draw inferences about long-run outcomes from contemporary behaviors. This is most clearly seen in the context of debates over economic convergence; as discussed below, many of the differences between neoclassical and endogenous growth perspectives may be reduced to questions concerning the long-run effects of initial conditions. The available growth data typically span at most 140 years (and many fewer if one wants to work with a data set that nontrivially spans countries outside Western Europe and the United States) and the use of these data to examine hypotheses about long-run behavior can be a difficult undertaking. Such exercises lead to complicated questions concerning how one can identify the steady-state behavior of a stochastic process from observations along its transition path. As we have already mentioned, another major and difficult set of growth questions involves the identification of empirically salient determinants of growth when the range of potential factors is large relative to the number of observations. Model uncertainty is in fact a fundamental problem facing growth researchers. Individual researchers, seeking to communicate the extent of support for particular growth determinants, typically emphasize a single model (or small set of models) and then carry out inference as if that model had generated the data. Standard inference procedures based on a single model, and which are conditional on the truth of that model, can grossly overstate the precision of inferences about a given phenomenon. Such procedures ignore the uncertainty that surrounds the validity of the model. Given that there are usually other models that have strong claims on our attention, the standard errors can understate the true degree of uncertainty about the parameters, and the choice of which models to report can appear arbitrary. The need to properly account for model uncertainty naturally leads to Bayesian or pseudo-Bayesian approaches to data analysis.2 Yet another set of questions involves the characterization of interesting patterns in a data set comprised of objects as complex and heterogeneous as countries. Assumptions about parameter constancy across units of observation seem particularly unappealing for cross-country data. On the other hand, much of the interest in growth economics stems precisely from the objective of understanding the distribution of outcomes across countries. The search for data patterns has led to a far greater use of classification and pattern 2 See Draper (1995) for a general discussion of model uncertainty and Brock, Durlauf and West (2003) for discussion of its implications for growth econometrics.
560
S.N. Durlauf et al.
recognition methods, for example, than appears in other areas of economics. Here and elsewhere, growth econometrics has imported a range of methods from statistics, rather than simply relying on the tools of mainstream econometrics. Whichever techniques are applied, the weakness of the available data represents a major constraint on the potential of empirical growth research. Perhaps the main obstacle to understanding growth is the small number of countries in the world. This is a problem for the obvious reason (a fundamental lack of variation or information) but also because it limits the extent to which researchers can address problems such as measurement error and parameter heterogeneity. Sometimes the problem is stark: imagine trying to infer the consequences of democracy for growth in poorer countries. Because the twentieth century provided relatively few examples of stable, multi-party democracies among the poorer nations of the world, statistical evidence can make only a limited contribution to this debate, unless one is willing to make exchangeability assumptions about nations that would seem not to be credible.3 With a larger group of countries to work with, many of the difficulties that face growth researchers could be addressed in ways that are now standard in the microeconometrics literature. For example, the well known concerns expressed by Harberger (1987), Solow (1994) and many others about assuming a common linear model for a set of very different countries could, in principle, be addressed by estimating more general models that allow for heterogeneity. This can be done using interaction terms, nonlinearities or semiparametric methods, so that the marginal effect of a given explanatory variable can differ across countries or over time. The problem is that these solutions will require large samples if the conclusions are to be robust. Similarly, some methods for addressing other problems, such as measurement error, are only useful in samples larger than those available to growth researchers. This helps to explain the need for new statistical methods for growth contexts, and why growth econometrics has evolved in such a pragmatic and eclectic fashion. One common response to the lack of cross-country variation has been to draw on variation in growth and other variables over time, primarily using panel data methods. Many empirical growth papers are now based on the estimation of dynamic panel data models with fixed effects. Our survey will discuss not only the relevant technical issues, but also some issues of interpretation that are raised by these studies, and especially their treatment of fixed effects as nuisance parameters. We also discuss the merits of alternatives. These include the before-and-after studies of specific events, such as stock market liberalizations or democratizations, which form an increasingly popular method for examining certain hypotheses. The correspondence between these studies and the microeconometric literature on treatment effects helps to clarify the strengths and limitations of the event-study approach, and of cross-country evidence more generally. Despite the many difficulties that arise in empirical growth research, we believe some progress has been made. Researchers have uncovered stylized facts that growth theories
3 See Temple (2000b) and Brock and Durlauf (2001a) for a conceptual discussion of this issue.
Ch. 8: Growth Econometrics
561
should endeavor to explain, and developed methods to investigate the links between these stylized facts and substantive economic arguments. We would also argue that an important contribution of growth econometrics has been the clarification of the limits that exist in employing statistical methods to address growth questions. One implication of these limits is that narrative and historical approaches [Landes (1998) and Mokyr (1992) are standard and valuable examples] have a lasting role to play in empirical growth analysis. This is unsurprising given the importance that many authors ascribe to political, social and cultural factors in growth, factors that often do not readily lend themselves to statistical analysis.4 For these reasons, Willard Quine’s classic statement of the underdetermination of theories by data, cited at the beginning of this chapter, seems especially relevant to the study of growth. The chapter is organized as follows. Section 2 describes a set of stylized facts concerning economic growth. These facts constitute the objects that formal statistical analysis has attempted to explain. Section 3 describes the relationship between theoretical growth models and econometric frameworks for growth, with a primary focus on cross-country growth regressions. Section 4 discusses the convergence hypothesis. Section 5 describes methods for identifying growth determinants, and a range of questions concerning model specification and evaluation are addressed. Section 6 discusses econometric issues that arise according to whether one is using cross-section, time series or panel data, and also examines the issue of endogeneity in some depth. Section 7 evaluates the implications of different data and error properties for growth analysis. Section 8 concludes with some thoughts on the progress made thus far, and possible directions for future research. 2. Stylized facts In this section we describe some of the major features of cross-country growth data. Our goal is to identify some of the salient cross-section and intertemporal patterns that have motivated the development of growth econometrics. Section 2.1 makes some general observations on growth in the very long-run. Section 2.2 discusses the main data set used to study growth since 1960. Section 2.3 describes general facts about differences in output per worker across countries. Section 2.4 extends this discussion by focusing on growth miracles and disasters. Basic facts concerning convergence are reported in Section 2.5. In Section 2.6 we describe the general slowdown in growth over the last two decades. Section 2.7 extends this discussion by considering the question of predictability of growth rates over time. Section 2.8 identifies growth differences across levels of development and across geographic regions. In Section 2.9, we characterize some aspects of stagnation and volatility. Section 2.10 draws some general conclusions about the basic growth facts. 4 Narrative approaches can, of course, be subjected to criticisms every bit as severe as apply to quantitative studies. Similarly, efforts to study qualitative growth ideas using formal tools can go awry; see Durlauf (2002) for criticism of efforts to explain growth and development using the idea of social capital.
562
S.N. Durlauf et al.
2.1. A long-run view Taking a long view of economic history, a central fact concerning aggregate economic activity across countries is the massive divergence in living standards that has occurred over the last several centuries. A snapshot of the world in 1700 would show all countries to be poor, if their living standards were assessed in today’s terms. Over the course of the 18th and 19th centuries, growth rates increased slightly in the UK and other countries in Western Europe. Annual growth rates appear to have remained low, by modern standards, even in the midst of the Industrial Revolution; but because this growth was sustained over time, GDP per capita steadily rose. The outcome was that the UK, some other countries in Western Europe, and then the USA gradually advanced further ahead of the rest of the world. What was happening elsewhere? As Pritchett (1997) argues, even in the absence of national accounts data, we can be almost certain that rapid productivity growth was never sustained in the poorer regions of the world. The argument proceeds by extrapolating backwards from their current levels of GDP per capita, using a fast growth rate. This quickly implies earlier levels of income that would be too low to support human life. 2.2. Data after 1960 Today’s overall inequality across countries is partly the legacy of rapid growth in a small group of Western economies, and its absence elsewhere. But there have been important deviations from this general pattern. Since the 1960s, some developing countries have grown at rates that are unprecedented, at least based on the experiences of the advanced economies of Europe and North America. The tiger economies of East Asia have seen GDP per worker grow at around 5% a year, or even faster, for the best part of forty years. A country that grows at such rates over forty years will see GDP per worker rise more than sevenfold, as in the case of Hong Kong, Singapore, South Korea and Taiwan. In the rest of this section, we describe these patterns in more detail. As with most of the empirical growth literature, we will focus on the period after 1960, the point at which national accounts data start to become available for a larger group of countries.5 Our calculations use version 6.1 of the Penn World Table (PWT) due to Heston, Summers and Aten (2002). They have constructed measures of real GDP that adjust for international differences in price levels, and are therefore more comparable across space than measures based on market exchange rates.6 For the purposes of our analysis, the “world” will consist of 102 countries, those with data available in PWT 6.1 and with populations of at least 350,000 in the year 1960. These 102 countries account for a large share of the world’s population. The
5 Another reason for this starting point is that many colonies did not gain independence until the 1960s. 6 For more discussion of the PWT data, and further references, see Temple (1999).
Ch. 8: Growth Econometrics
563
most important missing countries are economies in Eastern Europe that were centrally planned for much of the period. Because of its enormous population, collectivist China is included in the sample, but is a country for which output measurement is especially difficult. In a small number of cases, data for GDP per worker for 2000 are extrapolated from preceding years using growth rates for the early and mid-1990s. Appendix A gives more details of the sample, and the extrapolation procedure. Throughout, we use data on GDP per worker. Most formal growth models are based on production functions, and their implications relate more closely to GDP per worker than GDP per capita. Jones (1997) provides another justification for this choice. When there is an unmeasured non-market sector, such as subsistence agriculture, GDP per worker could be a more accurate index of average productivity than GDP per capita. The paths of GDP per worker and GDP per capita will diverge when there are changes in the ratio of workers to population, which is one form of participation rate. There has been an upwards trend in these participation rates where such rates were originally low, while at the upper end of the distribution participation has been stable.7 For a sample of 90 countries with available data, the median participation rate rose from 41% to 45% between 1960 and 2000. There was a sharp increase at the 25th percentile (from 33% to 40%) but very little change at the 75th percentile. This pattern suggests that growth in GDP per capita has usually been close to growth in GDP per worker, except for the countries that started with low participation rates. There is an important point to bear in mind, when interpreting our later tables and graphs, and those found elsewhere in the literature. Our unit of observation is the country. In one sense this is clearly an arbitrary way to divide the world’s population, but one that can have systematic effects on perceptions of stylized facts. We can illustrate this with a specific example. Sub-Saharan Africa has many countries that have small populations, while India and China combined account for about 40% of the world’s population. In a decade where India and China did relatively well, such as the 1990s, a country-based analysis will understate the overall improvement in living standards. In contrast, in a decade where Africa did relatively well, such as the 1960s, the overall growth record would appear less strong if assessed on a population-weighted basis. The point that countries differ greatly in terms of population size is important when interpreting tables, graphs and regressions that use the country as the unit of observation. 2.3. Differences in levels of GDP per worker Initially, we document the international disparities in GDP per worker. We first look at data for countries with large populations. Table 1 lists a set of countries that together account for 4.3 billion people. Of the countries with large populations, the main omissions are Germany, because of the difficulty posed by reunification, and economies that were centrally planned, including Russia.
7 The figures we use for participation rates are those implicit in the Penn World Table, 6.1.
564
S.N. Durlauf et al. Table 1 International disparities in GDP per worker Country USA United Kingdom Argentina France Italy South Africa Mexico Spain Iran Colombia Japan Brazil Turkey Philippines Egypt Korea, Republic of Bangladesh Nigeria Indonesia Thailand Pakistan India China Ethiopia Mean Median
Population (m, 2000)
R1960
R2000
275 60 37 60 58 43 97 40 64 42 127 170 67 76 64 47 131 127 210 61 138 1016 1259 64
1 0.69 0.62 0.60 0.55 0.47 0.44 0.40 0.30 0.27 0.25 0.24 0.17 0.17 0.17 0.15 0.10 0.08 0.08 0.07 0.07 0.06 0.04 0.04
1 0.69 0.40 0.76 0.84 0.34 0.38 0.68 0.30 0.18 0.60 0.30 0.24 0.13 0.21 0.57 0.10 0.02 0.14 0.20 0.11 0.10 0.10 0.02
0.29 0.21
0.35 0.27
Note: R is GDP per worker as a fraction of that in the USA.
The table shows GDP per worker, relative to the USA, for 1960 and 2000. The countries are ranked in descending order in terms of their 1960 position. Some clear patterns emerge: the major economies of Western Europe have maintained their position relative to the USA (as in the case of the UK) or substantially improved it (France, Italy, Spain). Among the poorer nations, there are some countries that have improved their relative position dramatically (Japan, Republic of Korea, Thailand) and others that have performed badly (Argentina, Nigeria). If we look at the mean and median of relative GDP per worker, there has been a moderate increase, suggesting a slight tendency for reduced dispersion. But these statistics disguise a wide variety of experience, and we will discuss the issue of convergence in more detail below. We now consider the shape of the international distribution of GDP per worker, using the USA’s 1960 value as the benchmark. Figure 1 shows a kernel density plot of the distribution of GDP per worker in 1960 and 2000, relative to the benchmark. The right-
Ch. 8: Growth Econometrics
565
Figure 1. Cross-country density of output per worker.
wards movement reflects the growth that took place over this period. Also noticeable is a thinning in the middle of the distribution, the “Twin Peaks” phenomenon identified in a series of papers by Quah (1993a, 1993b, 1996a, 1996b, 1996c, 1997). Is the position in the league table of GDP per worker in 1960 a good predictor of that in 2000? The answer is a qualified yes: the Spearman rank correlation is 0.84. This pattern is shown in more detail in Figure 2, which plots the log of GDP per worker relative to the USA in 2000, against that in 1960. In this and later figures, one or two outlying observations are omitted to facilitate graphing. The high rank correlation is not a new phenomenon. Easterly et al. (1993) report that, for 28 countries for which Maddison (1989) has data, the rank correlation of GDP per capita in 1988 with that in 1870 is 0.82. 2.4. Growth miracles and disasters Despite some stability in relative positions, it is easy to pick out countries that have done exceptionally well and others that have done badly. There is an enormous range in observed growth rates, to an extent that has not previously been observed in world history. To show this, we rank the countries by their annual growth rate between 1960 and 2000, and present a list of the fifteen best performers (Table 2) and the fifteen worst (Table 3). To show the dramatic effects of sustaining a high growth rate over forty years, we also show the ratio of GDP per worker in 2000 to that in 1960. These tables of growth miracles and disasters show a regional pattern that is familiar to anyone who has studied recent economic growth. The best performing countries are
566
S.N. Durlauf et al.
Figure 2. Output per worker: 1960 versus 2000. Table 2 Fifteen growth miracles, 1960–2000 Country Taiwan Botswana Hong Kong Korea, Republic of Singapore Thailand Cyprus Japan Ireland China Romania Mauritius Malaysia Portugal Indonesia
Growth 1960–2000
Factor increase
6.25 6.07 5.67 5.41 5.09 4.50 4.30 4.13 4.10 3.99 3.91 3.88 3.82 3.48 3.34
11.3 10.6 9.09 8.24 7.29 5.83 5.39 5.04 5.00 4.77 4.63 4.58 4.48 3.93 3.72
Ch. 8: Growth Econometrics
567 Table 3 Fifteen growth disasters, 1960–2000
Country Peru Mauritania Senegal Chad Mozambique Madagascar Zambia Mali Venezuela Niger Nigeria Nicaragua Central African Republic Angola Congo, Democratic Rep.
Growth 1960–2000
Ratio
0.00 −0.11 −0.26 −0.43 −0.50 −0.60 −0.61 −0.77 −0.88 −1.03 −1.21 −1.30 −1.56 −2.04 −4.00
1.00 0.96 0.90 0.84 0.82 0.79 0.78 0.74 0.70 0.66 0.62 0.59 0.53 0.44 0.20
mainly located in East Asia and Southeast Asia. These countries have sustained exceptionally high growth rates; for example, GDP per worker has grown by a factor of 11 in the case of Taiwan. If we now turn to the growth disasters, we can see many instances of “negative growth”, and these are predominantly countries in sub-Saharan Africa. Later in this section, we will compare Africa’s performance with that of other regions in more detail.8 2.5. Convergence? An alternative way of showing the diversity of experience is to plot the growth rate over 1960–2000 against the 1960 level of real GDP per worker, relative to the USA. This is shown in Figure 3. The most obvious lesson to be drawn from this figure is the diversity of growth rates, especially at low levels of development. The figure does not provide much support for the idea that countries are converging to a common level of income, since that would require evidence of a downward sloping relationship between growth and initial income. Neither does it support the widespread idea that poorer countries have always grown slowly. 2.6. The growth slowdown Next, we present similar figures for two sub-periods, 1960–1980 and 1980–2000. These plots, shown as Figures 4 and 5, reveal another important pattern. For many developing 8 Easterly and Levine (1997a, 1997b) and Collier and Gunning (1999a, 1999b) examine various explanations for slow growth in Africa.
568
S.N. Durlauf et al.
Figure 3. Growth versus initial income: 1960–2000.
countries, growth was significantly lower in the second period, with many countries seeing a decline in real GDP per worker after 1980. We can see this more clearly by looking at the international distribution of growth rates for the two sub-periods. Figure 6 shows kernel density estimates, and reveals a clear pattern: the mass of the distribution has shifted leftwards (slower growth) while at the same time the variance has increased (greater dispersion in growth rates). A different way to highlight the growth slowdown is to plot the growth rate in 1980– 2000 against that in 1960–1980 as is done in Figure 7, which also includes a 45 degree line. Countries above the line have seen growth increase, whereas countries below have seen growth decline. There are clearly more countries in which growth has declined over time, with the crucial exceptions of China and India, which have seen a dramatic improvement. To reveal the same pattern, Table 4 lists the countries in various categories, classified by growth rates in 1960–80 and in 1980–2000. 2.7. Does past growth predict future growth? Another lesson to be drawn from Figure 7 and Table 4 is that relative performance has been unstable. The correlation between growth in 1960–1980 and that in 1980–2000 is just 0.40, so past growth is not a particularly useful predictor of future growth.9 For
9 Easterly et al. (1993) emphasized this point, and suggested that the lack of persistence in growth rates indicates the importance of good luck.
Ch. 8: Growth Econometrics
569
Figure 4. Growth versus initial income 1960–1980.
Figure 5. Growth versus initial income: 1980–2000.
the whole sample, the correlations across decades are also weak (Table 5). It is less well known that the cross-decade correlation has tended to increase over time, as is clear from Table 5’s below diagonal elements for the whole sample. This is tentative evidence that
570
S.N. Durlauf et al.
Figure 6. Density of growth rates across countries.
Figure 7. Growth rates in 1960–1980 versus 1980–2000.
Ch. 8: Growth Econometrics
571 Table 4 Growth in 1960–1980 and 1980–2000
G2 0
0 < G2 1.5
1.5 < G2 3
G1 0
Angola, Central African Republic, DR Congo, Madagascar, Niger, Venezuela
Guinea, Mozambique, Senegal
0 < G1 1.5
Jamaica, Mali, Nicaragua, Nigeria, Rwanda, Zambia
Benin, El Salvador, Ethiopia, Guyana, New Zealand
Burkina Faso, Guinea-Bissau, Nepal, Sri Lanka
Bangladesh
1.5 < G1 3
Argentina, Bolivia, Burundi, Cameroon, Chad, Colombia, Costa Rica, Ghana, Honduras, Kenya, Papua New Guinea, Peru, Philippines, South Africa, Tanzania, Togo
Fiji, Gambia, Malawi, Mexico, Namibia, Netherlands, Sweden, Switzerland, Uruguay
Australia, Canada, Denmark, Chile, Dominican Rep., Egypt, Iran, Norway, UK, USA
China, India, Mauritius
G1 > 3
Ecuador, Gabon, Guatemala, Ivory Coast, Jordan, Mauritania, Panama, Paraguay, Zimbabwe
Brazil, Rep. Congo, France, Greece, Lesotho, Morocco, Spain, Syria, Trinidad and Tobago
Austria, Belgium, Finland, Indonesia, Israel, Italy, Japan, Pakistan, Portugal, Turkey
Botswana, Cyprus, Hong Kong, Ireland, Korea, Malaysia, Romania, Singapore, Taiwan, Thailand
G2 > 3 Uganda
Note: The above table classifies countries according to their annual growth rates over 1960–80 (G1) and over 1980–2000 (G2).
national economies are gradually sorting themselves into a pattern of distinct winners and losers. 2.8. Growth differences by development level and geographic region Can we say anything more about the characteristics of the winners and losers? First, we investigate the relationship between growth and initial development levels in more
572
S.N. Durlauf et al. Table 5 Growth rate correlations across decades 1960–1970
1970–1980
1980–1990
1990–2000
Whole sample Growth 1960–1970 Growth 1970–1980 Growth 1980–1990 Growth 1990–2000
1.00 0.16 0.28 0.11
1.00 0.31 0.33
1.00 0.44
1.00
Rich country group Growth 1960–1970 Growth 1970–1980 Growth 1980–1990 Growth 1990–2000
1.00 0.73 0.06 −0.07
1.00 0.40 0.37
1.00 0.61
1.00
Note: Whole sample is 102 countries. Rich country group is 19 countries.
Table 6 Growth, 1960–2000, by initial relative income Percentile All Relative income: R 0.05 R > 0.05 & R 0.10 R > 0.10 & R 0.25 R > 0.25 & R 0.50 R > 0.50
N
25th
Median
75th
102
0.7
1.6
2.7
10 22 33 19 18
1.0 −0.5 0.4 0.8 1.6
1.5 0.9 1.9 1.5 1.9
2.4 2.9 2.7 3.1 2.6
Notes: This table shows the 25th, 50th and 75th percentiles of the distribution of growth rates for countries at various levels of development in 1960. R is GDP per worker in 1960 relative to the US level.
detail. We rank the sample of 102 countries by initial income in 1960, and then look at the distribution of growth rates for subgroups. In Table 6, for various ranges of initial income relative to the USA, we show the growth rate at the 25th percentile, the median, and the 75th percentile. If we take the 22 countries which began somewhere between 5% and 10% of GDP per worker in the USA, the annual growth rate at the 25th percentile is negative, but is 2.9% at the 75th percentile. This diversity of experience extends throughout the distribution of relative incomes, but is less pronounced for the richest group.
Ch. 8: Growth Econometrics
573 Table 7 Growth, 1960–2000, by country groups
Group
N
25th
Median
75th
Sub-Saharan Africa South and Central America East and Southeast Asia South Asia Industrialized countries
36 21 10 7 19
−0.5 0.4 3.8 1.9 1.7
0.7 0.9 4.3 2.2 2.4
1.3 1.5 5.4 2.9 3.0
Note: This table shows the 25th, 50th and 75th percentiles of the distribution of growth rates for various groups of countries.
Table 7 shows the quartiles of growth rates for countries in different regions.10 Once again, sub-Saharan Africa is revealed as a weak performer. Within sub-Saharan Africa, even the country at the 75th percentile shows growth of just 1.3%. Performance is slightly better for South and Central America, but still not strong. Against this background, the record of East and Southeast Asia looks all the more remarkable. In further work (not shown) we have constructed versions of Tables 6 and 7 for 1960–1980 and 1980–2000. These reinforce the patterns already discussed: dispersion of growth rates at all levels of development, major differences across regional groups, and a collapse in growth rates after 1980. Even for the developed countries, growth rates were noticeably lower after 1980 than before, reflecting the well-known productivity slowdown and the reduced potential for catch-up by previously fast-growing countries, such as France, Italy and Japan. 2.9. Stagnation and output volatility Some countries did not record fast growth even in the boom of the 1960s. Some have simply stagnated or declined, never sustaining a high or even moderate growth rate for the length of time needed to raise output appreciably. In our sample, there are nine countries that have never exceeded their 1960 level of GDP per worker by more than 30%. Even more striking, a quarter of the countries (26 of 102) never exceeded their 1960 level by more than 60%. To put this in context, a country that grew at an average rate of 2% a year over a forty-year period would see GDP per worker rise by around 120%. Easterly (1994) drew attention to the international prevalence of stagnation, and the failure of some poorer countries to break out of low levels of development. There are other ways in which the behavior of the poorer countries looks very different to that of rich countries. As emphasized by Pritchett (2000a), it is not uncommon 10 These country groupings are not exhaustive; for example Fiji and Papua New Guinea do not appear in
any of these groups. Analysis of the group of industrialized countries is subject to the sample selection issue highlighted by DeLong (1988).
574
S.N. Durlauf et al. Table 8 Output collapses Country Chad Rwanda Angola Romania Dem. Rep. Congo Mauritania Tanzania Mali Cameroon Nigeria
Largest 3-year drop
Dates
50% 47% 46% 37% 36% 34% 34% 34% 33% 32%
1980–83 1991–94 1973–76 1977–80 1992–95 1985–88 1987–90 1985–88 1987–90 1997–00
Note: This table shows the ten countries with the largest output collapses over a three-year period, using data on GDP per worker between 1960 and the latest available year.
for output to undergo a major collapse in less developed countries (LDCs). To show this, we calculate the largest percentage drop in output over three years recorded for each country, using data from 1960 to the latest available year. The precise statistic we calculate is: Y2000 Y1963 Y1964 . , ,..., 100 · 1 − min Y1960 Y1961 Y1997 The largest ten output falls are shown in Table 8, which shows how dramatic an output collapse can be. Several of these output collapses are associated with periods of intense civil war, as in the cases of Rwanda, Angola and the Democratic Republic of the Congo. But the phenomenon of output collapse is a great deal more widespread than may be explained by events of this type. Of the 102 countries in our sample, 50 showed at least one three-year output collapse of 15% or more. 65 countries experienced a threeyear output collapse of 10% or more. In contrast, between 1960 and 2000, the largest three-year output collapse in the USA was 5.4%, and in the UK 3.6%, both recorded in 1979–82. A corollary of these patterns is that time series modeling of LDC output, whether on a country-by-country basis or using panel data, has to be approached with care. It is not clear that the dynamics of output in the wake of a major collapse would look anything like the dynamics at other times. We conclude our consideration of stylized facts by briefly reporting some evidence on long-run output volatility. Table 9 reports figures on the standard deviation of annual growth rates between 1960 and 2000. Industrialized countries are relatively stable, while sub-Saharan Africa is by far the most volatile region, followed by South and Central America. Volatility is not uniformly higher in developing countries, however: using the standard deviation of annual growth rates, South Africa is less volatile than the USA, Sri Lanka less volatile than Canada, and Pakistan less volatile than Switzerland.
Ch. 8: Growth Econometrics
575 Table 9 Volatility, 1960–2000, by regions
Group
N
25th
Median
75th
Sub-Saharan Africa South and Central America East and Southeast Asia South Asia Industrialized countries
36 21 10 7 19
5.5 3.9 3.8 3.0 2.3
7.4 4.8 4.1 3.3 2.9
9.3 5.4 4.7 5.2 3.5
Note: This table shows the 25th, 50th and 75th percentiles of the distribution of the standard deviation of annual growth rates, using data from the earliest available year until the latest available, between 1960 and 2000.
2.10. A summary of the stylized facts The stylized facts we consider can be summarized as follows: 1. Over the forty-year period as a whole, most countries have grown richer, but vast income disparities remain. For all but the richest group, growth rates have differed to an unprecedented extent, regardless of the initial level of development. 2. Although past growth is a surprisingly weak predictor of future growth, it is slowly becoming more accurate over time, and so distinct winners and losers are beginning to emerge. The strongest performers are located in East and Southeast Asia, which have sustained growth rates at unprecedented levels. The weakest performers are predominantly located in sub-Saharan Africa, where some countries have barely grown at all, or even become poorer. The record in South and Central America is also distinctly mixed. In these regions, output volatility is high, and dramatic output collapses are not uncommon. 3. For many countries, growth rates were lower in 1980–2000 than in 1960–1980, and this growth slowdown is observed throughout most of the income distribution. Moreover, the dispersion of growth rates has increased. A more optimistic reading would also emphasize the growth take-off that has taken place in China and India, home to two-fifths of the world’s population and a greater proportion of the world’s poor. Even this brief overview of the stylized facts reveals that there is much of interest to be investigated and understood. The field of growth econometrics has emerged through efforts to interpret and understand these facts in terms of simple statistical models, and in the light of predictions made by particular theoretical structures. In either case, the complexity of the growth process and the paucity of the available data combine to suggest that scientific standards of proof are unattainable. Perhaps the best this literature can hope for is to constrain what can legitimately be claimed. Researchers such as Levine and Renelt (1991) and Wacziarg (2002) have argued that, seen in this more modest light, growth econometrics can provide a signpost to interesting patterns and partial correlations, and even rule out some versions of the world that
576
S.N. Durlauf et al.
might otherwise seem plausible. Seen in terms of establishing stylized facts, empirical studies help to broaden the demands made of future theories, and can act as a discipline on quantitative investigations using calibrated models. In the remainder of this chapter, we will discuss in more detail the uses and limits of statistical evidence. We first examine how empirical growth studies are related to theoretical models, and then return in more depth to the study of convergence.
3. Cross-country growth regressions: from theory to empirics The stylized facts of economic growth have led to two major themes in the development of formal econometric analyses of growth. The first theme revolves around the question of convergence: are contemporary differences in aggregate economies transient over sufficiently long time horizons? The second theme concerns the identification of growth determinants: which factors seem to explain observed differences in growth? These questions are closely related in that each requires the specification of a statistical model of cross-country growth differences from which the effects on growth of various factors, including initial conditions, may be identified. In this section, we describe how statistical models of cross-country growth differences have been derived from theoretical growth models. Section 3.1 provides a general theoretical framework for understanding growth dynamics. The framework is explicitly neoclassical and represents the basis for most empirical growth work; even those studies that have attempted to produce evidence in favor of endogenous or other alternative growth theories have generally used the neoclassical model as a baseline from which to explore deviations. Section 3.2 examines the relationship between this theoretical model of growth dynamics and the specification of a growth regression. This transition from theory to econometrics produces the canonical cross-country growth regression. 3.1. Growth dynamics: basic ideas For economy i at time t, let Yi,t denote output, Li,t the labor force (assumed to obey Li,t = Li,0 eni t where the population growth rate ni is constant), and Ai,t the efficiency level of each worker with Ai,t = Ai,0 egi t where gi is the (constant) rate of (labor augmenting) technological progress. We will work with two main per capita notions: E = Y /(A L ) and output per labor unit output per efficiency unit of labor input, yi,t i,t i,t i,t yi,t = Yi,t /Li,t . As is well known, the generic one-sector growth model, in either its Solow–Swan or Ramsey–Cass–Koopmans variant, implies, to a first-order approximation, that E E E = 1 − e−λi t log yi,∞ + e−λi t log yi,0 , log yi,t (1) E is the steady-state value of y E and lim E E where yi,∞ t→∞ yi,t = yi,∞ . The parameter i,t E to its steady-state λi (which must be positive) measures the rate of convergence of yi,t
Ch. 8: Growth Econometrics
577
value and depends on the other parameters of the model. Given λi > 0, the value of E is independent of y E so that, in this sense, initial conditions do not matter in the yi,∞ i,0 long-run.11 E . In order to Equation (1) expresses growth dynamics in terms of the unobservable yi,t describe dynamics in terms of the observable variable yi,t we can write Equation (1) as E + e−λi t (log yi,0 − log Ai,0 ) log yi,t − gi t − log Ai,0 = 1 − e−λi t log yi,∞ (2) so that E log yi,t = gi t + 1 − e−λi t log yi,∞ + 1 − e−λi t log Ai,0 + e−λi t log yi,0 . In parallel to Equation (1), one can easily see that E lim yi,t − yi,∞ Ai,0 egi t = 0 t→∞
(3)
(4)
so that the initial value of output per worker has no implications for its long-run value. This description of the dynamics of output provides the basis for describing the dynamics of growth. Let γi = t −1 (log yi,t − log yi,0 )
(5)
denote the growth rate of output per worker between 0 and t. Subtracting log yi,0 from both sides of Equation (3) and dividing by t yields E − log Ai,0 , γi = gi + βi log yi,0 − log yi,∞ (6) where βi = −t −1 1 − e−λi t .
(7)
The βi parameter will prove to play a key role in empirical growth analysis. Equation (6) thus decomposes the growth rate in country i into two distinct components. The first component, gi , measures growth due to technological progress, whereas E − log A ) measures growth due to the gap the second component βi (log yi,0 − log yi,∞ i,0 between initial output per worker and the steady-state value, both measured in terms of efficiency units of labor. This second source of growth is what is meant by “catching up” in the literature. As t → ∞ the importance of the catch-up term, which reflects the role of initial conditions, diminishes to zero. Under the additional assumptions that the rates of technological progress, and the λi parameters are constant across countries, i.e. gi = g, and λi = λ ∀i, (6) may be rewritten as E γi = g − β log yi,∞ − β log Ai,0 + β log yi,0 .
(8)
11 Implicit in our discussion is the assumption that y E > 0 which eliminates the trivial equilibrium i,0 E = 0 ∀t. yi,t
578
S.N. Durlauf et al.
The important empirical implication of Equation (8) is that, in a cross-section of countries, we should observe a negative relationship between average rates of growth and initial levels of output over any time period – countries that start out below their balanced growth path must grow relatively quickly if they are to catch up with other countries that have the same levels of steady-state output per effective worker and initial efficiency. This is closely related to the hypothesis of conditional convergence, which is often understood to mean that countries converge to parallel growth paths, the levels of which are assumed to be a function of a small set of variables.12 Note, however, that a negative coefficient on initial income in a cross-country growth regression does not automatically imply conditional convergence in this sense, because countries might instead simply be moving toward their own different steady-state growth paths. 3.2. Cross-country growth regressions Equation (8) provides the motivation for the standard cross-country growth regression that is the foundation of the empirical growth literature. Typically, these regression specifications start with (8) and append a random error term υi so that E − β log Ai,0 + β log yi,0 + υi . γi = g − β log yi,∞
(9)
E and Implementation of (9) requires the development of empirical analogs for log yi,∞ log Ai,0 . Mankiw, Romer and Weil (1992) in a pioneering analysis, show how to do this in a way that produces a growth regression model that is linear in observable variables. In their analysis, aggregate output is assumed to obey a three-factor Cobb–Douglas production function φ
α Yi,t = Ki,t Hi,t (Ai,t Li,t )1−α−φ ,
(10)
where Ki,t denotes physical capital and Hi,t denotes human capital. Physical and human capital are assumed to follow the continuous time accumulation equations K˙ i,t = sK,i Yi,t − δKi,t
(11)
H˙ i,t = sH,i Yi,t − δHi,t
(12)
and
respectively, where δ denotes the depreciation rate, sK,i is the saving rate for physical capital, sH,i is the saving rate for human capital and dots above variables denote time derivatives. Note that the saving rates are both assumed to be time invariant. These accumulation equations, combined with the parameter constancy assumptions used to justify Equation (8) imply that the steady-state value of output per effective worker is 1 α sφ 1−α−φ sK,i H,i E yi,∞ = (13) α+φ (ni + g + δ) 12 We provide formal definitions of convergence in Section 4.1.
Ch. 8: Growth Econometrics
producing a cross-country growth regression of the form α+φ α log(ni + g + δ) − β log sK,i γi = g + β log yi,0 + β 1−α−φ 1−α−φ φ −β log sH,i − β log Ai,0 + υi . 1−α−φ
579
(14)
Mankiw, Romer and Weil assume that Ai,0 is unobservable and that g + δ is known. These assumptions mean that (14) is linear in the logs of various observable variables and therefore amenable to standard regression analysis. Mankiw, Romer and Weil argue that Ai,0 should be interpreted as reflecting not just technology, which they assume to be constant across countries, but country-specific influences on growth such as resource endowments, climate and institutions. They assume these differences vary randomly in the sense that log Ai,0 = log A + ei ,
(15)
where ei is a country-specific shock distributed independently of ni , sK,i , and sH,i .13 Substituting this into (14) and defining εi = υi − βei , we have the regression relationship α+φ log(ni + g + δ) γi = g − β log A + β log yi,0 + β 1−α−φ φ α log sK,i − β log sH,i + εi . −β (16) 1−α−φ 1−α−φ Using data from a group of 98 countries over the period 1960 to 1985, Mankiw, Romer and Weil produce regression estimates of βˆ = −0.299, αˆ = 0.48 and φˆ = 0.23.14,15 Mankiw, Romer and Weil are unable to reject the overidentifying restrictions present in (16). While this result is echoed in studies such as Knight, Loayza and Villanueva (1993), other authors, Caselli, Esquivel and Lefort (1996), for example, are able to reject the restrictions. Many cross-country regression studies have attempted to extend Mankiw, Romer and Weil by adding additional control variables Zi to the regression suggested by (16). Relative to Mankiw, Romer and Weil, such studies may be understood as allowing for predictable heterogeneneity in the steady-state growth term gi and initial technology term Ai,0 that are assumed constant across i in (16). Formally, the gi − β log Ai,0 terms 13 This independence assumption is justified, in turn, on the basis that (1) n , s i K,i , and sH,i are exogenous in
the neoclassical model with isoelastic preferences and (2) the estimated parameter values are consistent with those predicted by the model. 14 Based on data from the US and other economies, Mankiw, Romer and Weil set g + δ = 0.05 prior to estimation. 15 Using λ = −t −1 log(1 − tβ), the implied estimate of λ is 0.0142. The relationship λ = (1 − α − φ)(n + i i g + δ) was not imposed by Mankiw, Romer and Weil, who instead treat λ as a constant to be estimated. Durlauf and Johnson (1995, Table II, note b) show that estimating this model when λ varies with n in the way implied by the theory produces only very small changes in parameter estimates.
580
S.N. Durlauf et al.
in (6) are replaced with g − β log A + πZi − βei rather than with g − β log A − βei which produced (16). (As far as we know, empirical work universally ignores the fact that log(ni + g + δ) should also be replaced with log(ni + gi + δ).) This produces the cross country growth regression α+φ log(ni + g + δ) 1−α−φ φ −β log sH,i + πZi + εi . 1−α−φ
γi = g − β log A + β log yi,0 + β −β
α log sK,i 1−α−φ
(17)
The regression described by (17) does not identify whether the controls Zi are correlated with steady-state growth gi or the initial technology term Ai,0 . For this reason, a believer in a common steady-state growth rate will not be dissuaded by the finding that particular choices of Zi help predict growth beyond the Solow regressors. Nevertheless, it seems plausible that the controls Zi may sometimes function as proxies for predicting differences in efficiency growth gi rather than in the initial technology Ai,0 . As argued in Temple (1999), even if all countries have the same total factor productivity (TFP) growth in the long run, over a twenty- or thirty-year sample the assumption of equal TFP growth is highly implausible, so the variables in Zi can explain these differences. That being said, the attribution of the predictive content of Zi to initial technology versus steady state growth will entirely depend on a researcher’s prior beliefs. It is possible that proper accounting of the log(ni + gi + δ) term would allow for some progress in identifying gi versus Ai,0 effects since gi effects would imply a nonlinear relationship between Zi and overall growth γi ; however this nonlinearity may be too subtle to uncover given the relatively small data sets available to growth researchers. The canonical cross-country growth regression may be understood as a version of (17) when the cross-coefficient restrictions embedded in (17) are ignored (which is usually the case in empirical work). A generic representation of the regression is γi = β log yi,0 + ψXi + πZi + εi ,
(18)
where Xi contains a constant, log(ni + g + δ), log sK,i and log sH,i . The variables spanned by log yi,0 and Xi thus represent those growth determinants that are suggested by the Solow growth model whereas Zi represents those growth determinants that lie outside Solow’s original theory.16 The distinction between the Solow variables and Zi is important in understanding the empirical literature. While the Solow variables usually appear in different empirical studies, reflecting the treatment of the Solow model as a baseline for growth analysis, choices concerning which Zi variables to include vary greatly. Equation (18) represents the baseline for much of growth econometrics. These regressions are sometimes known as Barro regressions, given Barro’s extensive use of such 16 We distinguish log y i,0 from the other Solow variables because of the role it plays in analysis of conver-
gence; see Section 4 for detailed discussion.
Ch. 8: Growth Econometrics
581
regressions to study alternative growth determinants starting with Barro (1991). This regression model has been the workhorse of empirical growth research.17 In modern empirical analyses, the equation has been generalized in a number of dimensions. Some of these extensions reflect the application of (18) to time series and panel data settings. Other generalizations have introduced nonlinearities and parameter heterogeneity. We will discuss these variants below. 3.3. Interpreting errors in growth regressions Our development of the relationship between cross-country growth regressions and neoclassical growth theories illustrates the standard practice of adding regression errors in an ad hoc fashion. Put differently, researchers usually derive a deterministic growth relationship and append an error in order to capture whatever aspects of the growth process are omitted from the model that has been developed. One problem with this practice is that some types of errors have important implications for the asymptotics of estimators. Binder and Pesaran (1999) conduct an exhaustive study of this question, one important conclusion of which is that if one generalizes the assumption of a constant rate of technical change so that technical change follows a random walk, this induces nonstationarity in many levels series, raising attendant unit root questions. Beyond issues of asymptotics, the ad hoc treatment of regression errors leaves unanswered the question of what sorts of implicit substantive economic assumptions are made by a researcher who does this. Brock and Durlauf (2001a) address this issue using the concept of exchangeability. Basically, their argument is that in a regression such as (18), a researcher typically thinks of the errors εi as interchangeable across observations: different patterns of realized errors are equally likely to occur if the realizations are permuted across countries. In other words, the information available to a researcher about the countries is not informative about the error terms. Exchangeability is a mathematical formalization of this idea and is defined as follows. For each observation i, there exists an associated information set Fi available to the researcher. In the growth context, Fi may include knowledge of a country’s history or culture as well as any “economic” variables that are known. A definition of exchangeability (formally, F -conditional exchangeability) is µ(ε1 = a1 , . . . , εN = aN | F1 , . . . , FN ) = µ(ερ(1) = a1 , . . . , ερ(N ) = aN | F1 , . . . , FN ),
(19)
where µ( ) is a probability measure and ρ( ) is an operator that permutes the N indices. 17 Such regressions appear to have been employed earlier by Grier and Tullock (1989) and Kormendi and
Meguire (1985). The reason these latter two studies seem to have received less attention than warranted by their originality is, we suspect, due to their appearance before endogenous growth theory emerged as a primary area of macroeconomic research, in turn placing great interest on the empirical evaluation of growth theories. To be clear, Barro’s development is original to him and his linking of cross-country growth regressions to alternative growth theories was unique.
582
S.N. Durlauf et al.
Many criticisms of growth regressions amount to arguments that exchangeability has been violated. For example, omitted regressors induce exchangeability violations as these regressors are elements of F . Parameter heterogeneity also leads to nonexchangeability. For these cases, the failure of nonexchangeability calls into question the interpretation of the regression. This is not always the case; heteroskedasticity in errors violates exchangeability but does not induce interpretation problems for coefficients. Brock and Durlauf argue that exchangeability produces a link between substantive social science knowledge and error structure, i.e. this knowledge may be used to evaluate the plausibility of exchangeability. They suggest that a good empirical practice would be for researchers to question whether the errors in a model are exchangeable, and if not, determine whether the violation invalidates the purposes for which the regression is being used. This cannot be done in an algorithmic fashion, but as is the case with empirical work quite generally, requires judgments by the analyst. See Draper et al. (1993) for further discussion of the role of exchangeability in empirical work. 4. The convergence hypothesis Much of the empirical growth literature has focused on the convergence hypothesis. Although questions of convergence predate them, recent widespread interest in the convergence hypothesis originates from Abramovitz (1986) and Baumol (1986). This interest and the availability of the requisite data for a broad cross-section of countries, due to Summers and Heston (1988, 1991), spawned an enormous literature testing the convergence hypothesis in one or more of its various guises.18 In this section, we explore the convergence hypothesis. In Section 4.1 we consider the specification of notions of convergence as related to the relationship between initial conditions and long-run outcomes. Section 4.2 explores the main technique that has been employed in studying long-run dependence, β-convergence. Section 4.3 considers alternative notions of convergence that focus less on the persistence of initial conditions and instead on whether the cross-section dispersion of incomes is decreasing across time. This section explores σ -convergence, and more general notions as well as recent methods that fall under the heading of distributional dynamics. It also considers how distributional notions of convergence may be related to definitions found in Section 4.1. Section 4.4 develops time series approaches to convergence. Section 4.5 moves beyond the question of whether convergence is present to consider analyses that have attempted to identify the sources of convergence when it appears to be present. 4.1. Convergence and initial conditions The effect of initial conditions on long-run outcomes arguably represents the primary empirical question that has been explored by growth economists. The claim that the ef18 See Durlauf (1996) and the subsequent papers in the July 1996 Economic Journal, Durlauf and Quah
(1999), Islam (2003) and Barro and Sala-i-Martin (2004) for surveys of aspects of the convergence literature.
Ch. 8: Growth Econometrics
583
fects of initial conditions eventually disappear is the heuristic basis for what is known as the convergence hypothesis. The goal of this literature is to answer two questions concerning per capita income differences across countries (or other economic units, such as regions). First, are the observed cross-country differences in per capita incomes temporary or permanent? Second, if they are permanent, does that permanence reflect structural heterogeneity or the role of initial conditions in determining long-run outcomes? If the differences in per capita incomes are temporary, unconditional convergence (to a common long-run level) is occurring. If the differences are permanent solely because of cross-country structural heterogeneity, conditional convergence is occurring. If initial conditions determine, in part at least, long-run outcomes, and countries with similar initial conditions exhibit similar long-run outcomes, then one can speak of convergence clubs.19 We first consider how to formalize the idea that initial conditions matter. While the discussion focuses on log yi,t , the log level of per capita output in country i at time t; these definitions can in principle be applied to other variables such as real wages, life expectancy, etc. Our use of log yi,t rather than yi,t reflects the general interest in the growth literature in relative versus absolute inequality, i.e. one is usually more interested in whether the ratio of income between two countries exhibits persistence than an absolute difference, particularly since sustained economic growth will imply that a constant levels difference is of asymptotically negligible size when relative income is considered. We associate with log yi,t initial conditions, ρi,0 . These initial conditions do not matter in the long-run if lim µ(log yi,t | ρi,0 ) does not depend on ρi,0
t→∞
(20)
where µ(·) is a probability measure. To see how this definition connects with empirical growth work, note that empirical studies of convergence are often focused on whether long-run per capita output depends on initial stocks of human and physical capital. Economic interest in convergence stems from the question of whether certain initial conditions lead to persistent differences in per capita output between countries (or other economic units). One can thus use (20) to define convergence between two economies. Let denote a metric for computing the distance between probability measures.20 Then countries i and j exhibit convergence if lim µ(log yi,t | ρi,0 ) − µ(log yj,t | ρj,0 ) = 0. (21) t→∞
Growth economists are generally interested in average income levels; Equation (21) implies that countries i and j exhibit convergence in average income levels in the sense 19 This taxonomy is due to Galor (1996) who discusses the relationship between it and the theoretical growth
literature, giving several examples of models in which initial conditions matter for long-run outcomes. 20 There is no unique or single generally agreed upon metric for measuring deviations between probability
measures.
584
S.N. Durlauf et al.
that lim E(log yi,t − log yj,t | ρi,0 , ρj,0 ) = 0.
t→∞
(22)
To the extent one is interested in whether countries exhibit common steady-state growth rates, one can modify (22) to require that the limiting expected difference between log yi,t and log yj,t is bounded. One way of doing this is due to Pesaran (2004a) and is discussed below. These notions of convergence can be relaxed. Bernard and Durlauf (1996) suggest a form of partial convergence that relates to whether contemporaneous income differences are expected to diminish. If log yi,0 > log yj,0 , their definition amounts to asking whether E(log yi,t − log yj,t | ρi,0 , ρj,0 ) < log yi,0 − log yj,0 .
(23)
A number of modifications of these definitions have been proposed. Hall, Robertson and Wickens (1997) suggest appending a requirement that the variance of output differences diminish to 0 over time, i.e. lim E (log yi,t − log yj,t )2 | ρi,0 , ρj,0 = 0 (24) t→∞
so that convergence requires output for a pair of countries to behave similarly in the long-run. In our view, this is an excessively strong requirement since it does not allow one to regard the output series as stochastic in the long-run. Equation (24) would imply that convergence does not occur if countries are perpetually subjected to distinct business cycle shocks. However, Hall, Robertson and Wickens (1997) do identify a weakness of definition (22), namely the failure to control for long-run deviations whose current direction is not predictable. To see this, suppose that log yi,t − log yj,t is a random walk with current value 0. In this case, definition (22) would be fulfilled, although output deviations between countries i and j will become arbitrarily large at some future date. In recent work, Pesaran (2004a) has proposed a convergence definition that focuses specifically on the likelihood of large long-run deviations. Specifically, Pesaran defines convergence as lim Prob (log yi,t − log yj,t )2 < C 2 | ρi,0 , ρj,0 > π, (25) t→∞
where C denotes a deviation magnitude and π is a tolerance probability. The idea of this definition is to focus convergence analysis on output deviations that are economically important and to allow for some flexibility with respect to the probability with which they occur. These convergence definitions do not allow for the distinction between the long-run effects of initial conditions and the long-run effects of structural heterogeneity. From the perspective of growth theory, this is a serious limitation. For example, the distinctions between endogenous and neoclassical growth theories focus on the long-run
Ch. 8: Growth Econometrics
585
effects of cross-country differences initial human and physical capital stocks; in contrast, cross-country differences in preferences can have long-term effects under either theory. Hence, in empirical work, it is important to be able to distinguish between initial conditions ρi,0 and structural characteristics θi,0 . Steady state effects of initial conditions imply the existence of convergence clubs whereas steady-state effects of structural characteristics do not. In order to allow for this, one can modify (21) so that lim µ(log yi,t | ρi,0 , θi,0 ) − µ(log yj,t | ρj,0 , θj,0 ) = 0 if θi,0 = θj,0 (26) t→∞
implies that countries i and j exhibit convergence. The notions of convergence in expected value (Equation (22)) may be modified in this way as well, lim E(log yi,t − log yj,t | ρi,0 , θi,0 , ρj,0 , θj,0 ) = 0 if θi,0 = θj,0
t→∞
(27)
as can partial convergence in expected value (Equation (23)) and the other convergence concepts discussed above. In practice, the distinction between initial conditions and structural heterogeneity generally amounts to treating stocks of initial human and physical capital as the former and other variables as the latter. As such, both the Solow variables X and the control variables Z that appear in cross-country growth regression, cf. (18), are usually interpreted as capturing structural heterogeneity. This practice may be criticized if these variables are themselves endogenously determined by initial conditions, a point that will arise below. The translation of these ideas into restrictions on growth regressions has led to a range of statistical definitions of convergence which we now examine. Before doing so, we emphasize that none of these statistical definitions is necessarily of intrinsic interest per se; rather each concept is useful only to the extent it elucidates economically interesting notions of convergence such as Equation (20). The failure to distinguish between convergence as an economic concept and convergence as a statistical concept has led to a good deal of confusion in the growth literature. 4.2. β-convergence Statistical analyses of convergence have largely focused on the properties of β in regressions of the form (18). β-convergence, defined as β < 0 is easy to evaluate because it relies on the properties of a linear regression coefficient. It is also easy to interpret in the context of the Solow growth model, since the finding is consistent with the dynamics of the model. The economic intuition for this is simple. If two countries have common steady-state determinants and are converging to a common balanced growth path, the country that begins with a relatively low level of initial income per capita has a lower capital–labor ratio and hence a higher marginal product of capital; a given rate of investment then translates into relatively fast growth for the poorer country. In turn, β-convergence is commonly interpreted as evidence against endogenous growth models of the type studied by Romer and Lucas, since a number of these models specifically
586
S.N. Durlauf et al.
predict that high initial income countries will grow faster than low initial income countries, once differences in saving rates and population growth rates have been accounted for. However, not all endogenous growth models imply an absence of β-convergence and therefore caution must be exercised in drawing inferences about the nature of the growth process from the results of β-convergence tests.21 There now exists a large body of studies of β-convergence, studies that are differentiated by country set, time period and choice of control variables. When controls are absent, β < 0 is known as unconditional β-convergence: conditional β-convergence is said to hold if β < 0 when controls are present. Interest in unconditional β-convergence, while not predicted by the Solow growth model except when countries have common steady-state output levels, derives from interest in the hypothesis that all countries are converging to the same growth path, which is critical in understanding the extent to which current international inequality will persist into the far future.22 Typically, the unconditional β-convergence hypothesis is supported when applied to data from relatively homogeneous groups of economic units such as the states of the US, the OECD, or the regions of Europe; in contrast there is generally no correlation between initial income and growth for data taken from more heterogeneous groups such as a broad sample of countries of the world.23 Many cross-section studies employing the β-convergence approach find estimated convergence rates of about 2% per year.24 This result is found in data from such diverse entities as the countries of the world (after the addition of conditioning variables), the OECD countries, the US states, the Swedish counties, the Japanese prefectures, the regions of Europe, the Canadian provinces, and the Australian states, among others; it is also found in data sets that range over time periods from the 1860’s though the 1990’s.25 Some writings go so far as to give this value a status analogous to a universal 21 Jones and Manuelli (1990) and Kelly (1992) are early examples of endogenous growth models compatible
with β-convergence. Each model produces steady state growth without exogenous technical change yet each implies relatively fast growth for initially capital poor economies. 22 Formally, β-convergence is an implication of (9) if log y E is assumed constant across countries in addii,∞ tion to the assumption on log Ai,0 made in (15). 23 See Barro and Sala-i-Martin (2004, Chapters 11 and 12) for application of β-convergence tests to a variety of data sets. Homogeneity can reflect self-selection as pointed out by DeLong (1988). He argues that Baumol’s (1986) conclusion that unconditional β-convergence occurred over 1870–1979 among a set of affluent (in 1979) countries is spurious for this reason. 24 Panel studies estimates of convergence rates have typically been substantially higher than cross-section estimates. Examples where this is true for regressions that only control for the Solow variables include Islam (1995) and Lee, Pesaran and Smith (1998). The panel approach has possible interpretation problems which we discuss in Section 6. 25 For example, Barro and Sala-i-Martin (1991) present results for US states and regions as well as European regions; Barro and Sala-i-Martin (1992) for US states, a group of 98 countries and the OECD; Mankiw, Romer and Weil (1992) for several large groups of countries; Sala-i-Martin (1996a, 1996b) for US states, Japanese prefectures, European regions, and Canadian provinces; Cashin (1995) for Australian states and New Zealand; Cashin and Sahay (1996) for Indian regions; Persson (1997) for Swedish counties; and, Shioji (2001a) for Japanese prefectures and other geographic units.
Ch. 8: Growth Econometrics
587
constant in physics.26 In fact, there is some variation in estimated convergence rates, but the range is relatively small; estimates generally range between 1% and 3%, as noted by Barro and Sala-i-Martin (1992).27 Despite the many confirmations of this result now in the literature, the claim of global conditional β-convergence remains controversial; here we review the primary problems with the β-convergence literature. 4.2.1. Robustness with respect to choice of control variables In moving from unconditional to conditional β-convergence, complexities arise in terms of the specification of steady-state income. The reason for this is the dependence of the steady-state on Z. Theory is not always a good guide in the choice of elements of Z; differences in formulations of Equation (18) have led to a “growth regression industry” as researchers have added plausibly relevant variables to the baseline Solow specification. As a result, one can identify variants of (18) where convergence appears to occur as βˆ < 0 as well as variants where divergence occurs, i.e. βˆ > 0. We discuss issues of uncertainty in the specification of growth regressions below. Here we note here that one class of efforts to address model uncertainty has led to confirmatory evidence of conditional β-convergence. This approach assigns probabilities to alternative formulations of (18) and uses these probabilities to construct statements about β that average across the different models. Doppelhofer, Miller and Sala-i-Martin (2004) conclude the posterior probability that initial income is part of the linear growth model is 1.00 with a posterior expected value for β of −0.013; this leads to a point estimate of a convergence rate of 1.3% per annum, which is somewhat lower than the 2% touted in the literature; Fernandez, Ley and Steel (2001a) also find that the posterior probability that initial income is part of the linear growth model is 1.00, despite using a different set of potential models and different priors on model parameters.28 We therefore conclude that the evidence for conditional β-convergence appears to be robust with respect to choice of controls. 26 An alternative view is expressed by Quah (1996b) who suggests that the 2% finding may be a statistical ar-
tifact that arises for reasons unrelated to convergence per se. At the most primitive level, like any endogenous variable, the rate of convergence is determined by preferences, technology, and endowments. Operationally, this means that the rate of convergence will depend on model parameters and exogenous variables. For example, as stated above, in the augmented Solow model studied by Mankiw, Romer and Weil (1992), the relationship between the rate of convergence and the parameters of the model is λi = (1 − α − φ)(ni + g + δ). Barro and Sala-i-Martin (2004, pp. 111–113) discuss the relationship for the case of the Ramsey–Cass– Koopmans model with an isoelastic utility function and a Cobb–Douglas production function. Given this dependence, the ubiquity of the estimated 2% rate of convergence, taken at face value, appears to suggest a remarkable uniformity of preferences, technologies, and endowments across the economic units studied. 27 Barro and Sala-i-Martin argue that this variation reflects unobserved heterogeneity in steady-state values with more variation being associated with slower convergence. However, in as much as it is correlated with variables included in the regression equations, unobserved heterogeneity renders the parameter estimators inconsistent, which renders the estimated convergence parameter hard to interpret. 28 Fernandez, Ley and Steel (2001a) do not report a posterior expected value for β.
588
S.N. Durlauf et al.
4.2.2. Identification and nonlinearity: β-convergence and economic divergence A second problem with the β-convergence literature is an absence of attention to the relationship between β-convergence and economic convergence as defined by Equation (20) or variations based upon it. Put differently, in the β-convergence literature there is a general failure to develop tests of the convergence hypothesis that discriminate between convergent economic models and a rich enough set of non-converging alternatives. While β < 0 is an implication of the Solow growth model and so is an implication of the baseline convergent growth model in the literature, this does not mean that β < 0 is inconsistent with economically interesting non-converging alternatives. One such example is the model of threshold externalities and growth developed by Azariadis and Drazen (1990). In this model, there is a discontinuity in the aggregate production function for aggregate economies. This discontinuity means that the steady-state behavior of a given economy depends on whether its initial capital stock is above or below this threshold; specifically, this model may exhibit two distinct steady states. (Of course, there can be any number of such thresholds.) An important feature of the AzariadisDrazen model is that data generated by economies that are described by it can exhibit statistical convergence even when multiple steady states are present. To illustrate this, we follow an argument in Bernard and Durlauf (1996) based on a simplified growth regression. Suppose that for every country in the sample, the Solow variables Xi and additional controls Zi are identical. Suppose as well that there is no technical change or population growth. Following the standard arguments for deriving a cross-country regression specification, the growth regression implied by the Azariadis– Drazen assumption on the aggregate production function is ∗ γi = k + β log yi,0 − log yl(i) + εi ,
(28)
∗ denotes where l(i) indicates the steady state with which country i is associated and yl(i) output per capita in that steady state; all countries associated with the same steady state ∗ value. thus have the same log yl(i) The threshold externality model clearly does not exhibit economic convergence as defined above so long as there are at least two steady states. Yet the data generated by a cross-section of countries exhibiting multiple steady states may exhibit statistical convergence. To see this, notice that for this stylized case, the cross-country growth regression may be written as
γi = k + β log yi,0 + εi .
(29)
Since the data under study are generated by (28), this standard regression is misspecified. What happens when (29) is estimated when (28) is the data generating process? Using population moments, the estimated convergence parameter βols will equal
Ch. 8: Growth Econometrics
βols = β
589
∗ ), log y ) cov((log yi,0 − log yl(i) i,0
var(log yi,0 ) ∗ , log y ) cov(log yl(i) i,0 =β 1− . var(log yi,0 )
(30)
From the perspective of tests of the convergence hypothesis, the noteworthy feature of (30) is that one cannot determine the sign of βols a priori as it depends on ∗ , log y )/var(log y ), which is a function of the covariance between 1 − cov(log yl(i) i,0 i,0 the initial and steady-state incomes of the countries in the sample. It is easy to see that it is possible for βols to be negative even when the sample includes countries associated with different steady states. Roughly speaking, one would expect βols < 0 if low-income countries tend to initially be below their steady states whereas high-income countries tend to start above their steady states. While we do not claim this is necessarily the case empirically, the example does illustrate how statistical convergence (defined as β < 0) may be consistent with economic nonconvergence. Interestingly, it is even possible for the estimated convergence parameter βols to be smaller (and hence imply more rapid convergence) than the structural parameter β in (28). Below, we review evidence of multiple steady states in the growth process. At this stage, we would note two things. First, some studies have produced evidence of multiple regimes in the sense that statistical models consistent with multiple steady states appear to better fit the cross-country data than the linear Solow model, e.g., Durlauf and Johnson (1995). Second, other studies have produced evidence of parameter heterogeneity such that β appears to depend nonlinearly on initial conditions so that it is equal to 0 for some countries; Liu and Stengos (1999) find precisely this when they reject the specification of constant β for all countries in favor of a specification in which β depends on initial income. These types of findings imply the compatibility of observed growth patterns with the existence of permanent income differences between economies with identical population growth and savings rates and access to identical technologies. 4.2.3. Endogeneity A third criticism that is sometimes made of the empirical convergence literature is based on the failure to account for the endogeneity of the explanatory regressors in growth regressions. One obvious reason why endogeneity may matter concerns the consistency of the regression estimates. This concern has led some authors to propose instrumental variables approaches to estimating β. Barro and Lee (1994) analyze growth data in the periods 1965 to 1975 and 1975 to 1985 and use 5-year lagged explanatory variables as instruments. Barro and Lee find that the use of instrumental variables has little effect on coefficient estimates. Caselli, Esquivel and Lefort (1996) employ a generalized method of moments (GMM) estimator to analyze a panel variant of the standard cross-country growth regression; growth in the panel is measured in 5-year intervals for 1960–1985. Their analysis produces estimates of β on the order of 10%, which is much larger than the 2% typically found.
590
S.N. Durlauf et al.
Endogeneity raises a second identification issue with respect to the relationship between β-convergence and economic convergence: this idea appears in Cohen (1996) and Goetz and Hu (1996). Focusing on the Solow regressors, the value of β can fail to illustrate how initial conditions affect expected future income differences if the population and saving rates are themselves functions of income. Hence, β 0 may be compatible with at least partial economic convergence, if the physical and human capital savings rates depend, for example, on the level of income. In contrast, β < 0 may be compatible with economic divergence if the physical and human capital accumulation rates for rich and poor are diverging across time. As such, this critique is probably best understood as a debate over what variables are the relevant initial conditions for evaluating (22) and/or (23). Cohen (1996) argues that the conventional human capital accumulation equation, in which accumulation is proportional to per capita output, is misspecified, failing to account for feedbacks from the stock of human capital to the accumulation process. This feedback means that poor countries with low initial stocks of human capital fail to accumulate human capital as quickly as richer ones. Goetz and Hu (1996) directly focus on the feedback from income to human capital accumulation. The implications of this form of endogeneity for empirical work on convergence are mixed. Cohen (1996) concludes that a proper accounting for the dependence of human capital accumulation on initial capital stocks reconciles conditional β-convergence with unconditional β-divergence for a broad cross-section. Goetz and Hu (1996), in contrast, find that estimates of the speed of convergence are increased if one accounts for the effect of income on human capital accumulation for counties in the US South. This seems to be an area that warrants much more work. 4.2.4. Measurement error As Abramovitz (1986), Baumol (1986), DeLong (1988), Romer (1990), and Temple (1998) point out, measurement errors will tend to bias regression tests towards results consistent with the hypothesis of β-convergence. This occurs because, by construction, γi,t is measured with positive (negative) error when log yi,0 is measured with negative (positive) error so there tends to be a negative correlation between the measured values of the two variables even if there is none between the true values. To see this, we ignore the issue of control variables and consider the case where growth is described by γi = k + β log yi,0 + εi where εi is independent across observations. Suppose that log output is measured with error so that the researcher only observes ςi,t = log yi,t +ei,t , t = 0, T where ei,t is a serially uncorrelated random variable with variance σe2 and distributed independently of log yi,s and εi for all i and s. The regression of observed growth rates will, under these assumptions, obey the equation βT + 1 −1 −1 T (ςi,T − ςi,0 ) = k + βςi,0 + T ei,T − (31) ei,0 + εi . T This is a classic errors in variables problem; the term ( βTT+1 )ei,0 is negatively correlated ˆ In other words, the regression with ςi,0 which induces a negative bias in the estimate β.
Ch. 8: Growth Econometrics
591
of observed growth rates on observed initial incomes will tend to produce an estimated coefficient that is consistent with the β-convergence hypothesis even if the hypothesis is not reflected in the actual behavior of growth rates across countries. In practice, as Temple (1998) explains, the direction of the bias is made ambiguous by the possibilities that the ei,t are serially dependent and that other right-hand-side (conditioning) variables are also measured with error. The actual effect of measurement error on results then becomes an empirical matter to be investigated by individual researchers. In studying the role of the level of human capital in determining the rate of growth, Romer (1990) estimates a growth equation that has among its explanatory variables the level of per capita income at the beginning of the sample period. Consistent with the conditional β-convergence hypothesis, he finds a negative and significant coefficient on this variable when the equation is estimated by ordinary least squares. Wary of the possibility and effects of measurement error in initial income, as well as in the human capital variable – the literacy rate – Romer also estimates the equation using the number of radios per 1000 inhabitants and (the log of) per capita newsprint consumption as instruments for initial income and literacy with the result that the coefficients on both variables become insignificant “suggesting” that the OLS results are “attributable to measurement error” (p. 278). Temple (1998) uses the measurement error diagnostics developed by Klepper and Leamer (1984) and Klepper (1988), in addition to classical method-of-moments adjustments, to investigate the effects of measurement error on the estimated rate of convergence in Mankiw, Romer and Weil augmented Solow model. He finds that allowing for the possibility of small amounts of unreliability in the measurement of initial income implies a lower bound on the estimated convergence rate just above zero – too low to elevate conditional convergence to the status of a stylized fact. Barro and Sala-i-Martin (2004, pp. 472–473) use lagged values of state personal income as instruments for initial income to check for the possible effects of measurement error in their β-convergence tests for the US states. They find little change in the estimated convergence rates and conclude that measurement error is not an important determinant of their results. Barro (1991) follows the same procedure for other data sets and reaches a similar conclusion about the unimportance of measurement error in his results. Some authors have attempted to address the sources of measurement error. Dowrick and Quiggin (1997) is a notable example in this regard in their consideration of the role of price indices in affecting convergence tests. Specifically, they examine the effect of constant price estimates of GDP on β-convergence calculations and find that when the prices used to construct these measures are based on prices in advanced economies, tendencies towards convergence are understated. 4.2.5. Effects of linear approximation There is a body of research that explores the effects of the approximations that are employed to produce the linear regression models used to evaluate β-convergence. As outlined earlier, regression tests of the β-convergence hypothesis rely on a log-linear
592
S.N. Durlauf et al.
approximation to the law of motion in a one sector neoclassical growth model. In addition to the possibility that Taylor series approximations in the nonstochastic version of the model are inadequate, Binder and Pesaran (1999) show that the standard practice of adding a random term to the log-linearized solution of a nonstochastic growth model does not necessarily produce the same behavior as associated with the explicit solution of a stochastic model. Efforts to explore the limits of the linear approximation used in empirical growth studies have generally concluded that the approximation is reasonably accurate. Romer (2001, p. 25, n. 18) claims that the approximation will be “quite reliable” in this context and Dowrick (2004) presents results showing that the approximation to the true transition dynamics is quite good in a Solow model with a single capital good and an elasticity of output with respect to capital of 2/3. This is larger than the typical physical capital share but it is not an unreasonable number for the sum of the shares of physical and human capital. To test for nonlinearity, Barro (1991) adds the square of initial (1960) income to one of his regressions and finds a positive estimated coefficient implying that the rate of convergence declines as income rises and that it is positive only for incomes below $10800 – a figure that exceeds all of the 1960 income levels in his sample. However, the t-ratio for the estimated coefficient on the square of initial income is just 1.4 which represents weak evidence against the adequacy of the approximation. How should one interpret such findings? At one level, these studies conclude that the approximation used to derive the equation used in cross-section convergence studies appears to be reasonably accurate. It follows that the previously discussed nonlinearities in the growth process found by researchers investigating the possibility of multiple steady states do not reflect the inadequacy of the linear approximation used in most cross-section studies. Put differently, evidence of nonlinearity appears to reflect deeper factors than simple approximation error from the use of a first order Taylor series expansion. 4.3. Distributional approaches to convergence A second approach to convergence focuses on the behavior of the cross-section distribution of income in levels. Unlike the β-convergence approach, the focus of this literature has been less on the question of relative locations within the income distribution, i.e. whether one can expect currently poor countries to either equal or exceed currently affluent countries, but rather on the shape of the distribution as a whole. Questions of this type naturally arise in microeconomic analyses of income inequality, in which one may be concerned with whether the gap between rich and poor is diminishing, regardless of whether the relative positions of individuals are fixed over time. 4.3.1. σ -convergence Much of the empirical literature on the cross-country income distribution has focused on the question of the evolution of the cross-section variance of log yi,t . For a set of
Ch. 8: Growth Econometrics
593
2 income levels let σlog y,t denote the variance across i of log yi,t . σ -convergence is said to hold between times t and t + T if 2 2 σlog y,t − σlog y,t+T > 0.
(32)
This definition is designed, like β-convergence, to formalize the idea that contemporary income differences are transitory, but does so by asking whether the dispersion of these differences will decline across time. Recent work has attempted to identify regression specifications from which one can infer σ -convergence. Friedman (1992) and Cannon and Duck (2000) argue that it is possible to produce evidence concerning σ -convergence from regressions of the form γi = T −1 (log yi,t+T − log yi,t ) = α + π log yi,t+T + εi .
(33)
To see why this is so, following Cannon and Duck (2000), observe that σ -convergence 2 requires that σlog yi,t ,log yi,t+T < σlog yi,t . The regression coefficient in (33) may be written as σlog yi,t ,log yi,t+T π = T −1 1 − (34) 2 σlog yi,t+T 2 which means that π < 0 implies σlog yi,t ,log yi,t+T < σlog yi,t . Positiveness definiteness of
the variance/covariance matrix for log yi,t and log yi,t+T requires that (σlog yi,t ,log yi,t+T )2 2 2 < σlog yi,t σlog yi,t+T . Therefore, if π < 0, then it must be the case that (32) holds. Hence a test that accepts the null hypothesis that π < 0 by implication accepts the null hypothesis of σ -convergence. But even this type of test has some difficulties. As pointed out by Bliss (1999, 2000), it is difficult to interpret tests of σ -convergence since these tests presume that the data generating process is not invariant; an evolving distribution for the data makes it difficult to think about test distributions under a null. Additional issues arise when unit roots are present. One limitation to this approach is that it is not clear how one can formulate a sensible notion of conditional σ -convergence. A particular problem in this regard is that one would not want to control for initial income in forming residuals, which would render the concept uninteresting as it could be generated by nothing more than time-dependent heteroskedasticity in the residuals. On the other hand, omitting income would render the interpretation of the projection residuals problematic since initial income is almost certain to be correlated with the variables that have been included when the residuals are formed. An economically interesting formulation of conditional σ -convergence would be a useful contribution. 4.3.2. Evolution of the world income distribution Work on σ -convergence has helped stimulate the more general study of the evolution of the world income distribution. This work involves examining the cross-section distribution of country incomes at two or more points in time in order to identify how
594
S.N. Durlauf et al.
this cross-section distribution has changed. Of particular interest in such studies is the presence or emergence of multiple modes in the distribution. Bianchi (1997) uses nonparametric methods to estimate the shape of the cross-country income distribution and to test for multiple modes in the estimated density. He finds evidence of two modes in densities estimated for 1970, 1980, and 1989. Moreover, he finds a tendency for the modes to become more pronounced and to move further apart over time. This evidence supports the ideas of a vanishing middle as the distribution becomes increasingly polarized into “rich” and “poor” and of a growing disparity between those two groups. While such polarization might be desirable, were it the case that middle income economies were becoming high income ones, Bianchi’s evidence suggests that much of this movement represents a transition from middle income to poor. Further, by “cutting” each of the estimated densities at the anti-mode between the two modes, Bianchi is able to measure mobility within the distribution by counting the crossings of the cut points. These crossings represent countries moving from one basin of attraction to the other. Just 3 of the possible 238 crossings are observed.29 The implication is that there is very little mobility within the cross-country income distribution. The 20 or so countries in the “rich” basin of attraction in 1970 are still there in 1989 and similarly for the 100 or so countries starting in the “poor” basin. Paap and van Dijk (1998) model the cross-country distribution of per capita income as the mixture of a Weibull and a truncated normal density. The Weibull portion captures the left-hand mode and right skewness in the data while the truncated normal portion captures the right-hand mode. This combination is selected after testing the goodness of fit of various combinations of the normal density (truncated at zero), gamma, log normal and Weibull distributions; the data set that is employed measures levels of real GDP per capita for 120 countries for the time period 1960 and 1989. They find a bimodal fitted density in each year with “poor” and “rich” components corresponding to the Weibull and truncated normal densities respectively. The computed means of these components diverge over the sample period and the weight given to the poor component in the mixture jumps in the mid-1970’s from about 0.72 to about 0.82 implying that the mean gap between rich and poor countries grew and the poor increased in number. The attention to levels rather than log levels makes it hard to evaluate the welfare significance of this increased dispersion. Recently, analyses of the distributions of income and growth have focused on identifying differences in these distributions across time and across subsets of countries. Anderson (2003) studies changes in the world income distribution by using nonparametric density function estimates combined with stochastic dominance arguments to
29 Bianchi’s data contains 119 countries observed at 3 distinct years, so each country is capable of making
two crossings. The only crossings observed are (1) Trinidad and Tobago, which moves down between 1980 and 1989, (2) Venezuela, which moves down between 1970 and 1980, and (3) Hong Kong, which moves up between 1970 and 1980.
Ch. 8: Growth Econometrics
595
compare the distributions at different points in time.30 These methods allow him to construct measures of polarization of the income distribution; polarization is essentially characterized by shifts in probability density mass that increase disparities between relatively rich and relatively poor economies. Anderson finds that between 1970 and 1995 polarization between rich and poor countries increased throughout the time period. Maasoumi, Racine and Stengos (2003) analyze the evolution of the cross-country distributions of realized, predicted, and residual growth rates; fitted growth rates and residuals are formed from nonparametric growth regressions using the Solow variables. These authors find that the distributions of growth rates for OECD and non-OECD countries are persistently different between 1965 and 1995, with the OECD distribution’s variance reducing over time whereas the non-OECD distribution appears to be becoming less concentrated. One finds the same results for fitted growth rates; in contrast it is difficult to identify dimensions along which the distributions of OECD and non-OECD growth rate residuals differ. The major methodological difference between these papers relative to Paap and van Dijk (1998) is that these analyses do not rely on a mixture specification. Distributional approaches suggest the utility of convergence measures that are based on the complete properties of probability measures characterizing output for different economies. Letting µi (x) and µj (x) denote the probability density functions for the variable of interest in economies i and j respectively, Anderson and Ge (2004) propose computing the convergence statistic CIi,j CIi,j =
∞ −∞
min µi (x), µj (x) dx.
(35)
This statistic is bounded between 0 and 1; a value of zero means that the density functions never assign positive probability to any common intervals or values of x whereas a value of 1 means that the densities coincide on all positive probability intervals or values. Anderson and Ge (2004) refer to the case CIi,j = 1 as complete convergence. This statistic differs from the convergence measure described by Equation (21) as it evaluates differences between current densities and not asymptotic ones, but they are clearly closely related. In our view, this approach will likely prove useful in a range of contexts. In particular, if one is interested in comparing income distributions between two economies, the Anderson–Ge statistic is a natural metric. In growth contexts, it is less clear whether the higher moments that distinguish (22) from (35) are of major concern, at least in the context of current debates.
30 Anderson (2004) discusses issues related to the interpretation and econometric implementation of these
methods.
596
S.N. Durlauf et al.
4.3.3. Distribution dynamics In a series of papers, Quah (1993a, 1993b, 1996a, 1996b, 1996c, 1997) has persuasively criticized standard regression approaches to studying convergence issues for being unable to shed light on important issues of mobility, stratification, and polarization in the world income distribution. Rather than studying the average behavior of a representative country, Quah proposes a schema, which he calls “distribution dynamics”, for studying the evolution of the entire cross-country income distribution. One way of implementing this approach is to assume that the process describing the evolution of the distribution is time-invariant and first-order Markov. Discretizing the state space then permits representation of the cross-country income distribution as a probability mass function, λt , with an associated transition matrix, M. Each row of M is a probability mass function describing the distribution over states of the system after one transition given that the system is currently in the state corresponding to that row. The evolution of the income distribution can then be described by λt = M λt−1 so that λt+s = (M s ) λt is the sstep-ahead probability mass function and λ∞ = M λ∞ defines the long-run (ergodic) mass function (if it exists). Quah (1993b, 1996b) takes this approach and finds that the estimated M implies a bimodal (“twin-peaked”) ergodic mass function indicating a tendency towards polarization in the evolution of the world income distribution.31 Updating Quah’s analysis using more recent data, Kremer, Onatski and Stock (2001) also find evidence of twin-peaks in the long-run distribution of per capita incomes. However, they find the rich (right-hand) peak to be much larger than the poor (left-hand) peak unlike Quah, who found similarly sized peaks at both ends of the distribution. Kremer, Onatski and Stock’s point estimates imply that most countries will ultimately move to the rich state although, during the transition period, which could last hundreds of years, polarization in the income distribution may worsen. They are also unable to reject the hypothesis that there is a single right-hand peak in the long-run distribution. Quah (2001) responds to these claims by arguing that the imprecision in the estimates of the ergodic distributions is such that it is not possible to reject a wide range of null hypotheses including, by construction, that of twin-peakedness. Importantly, as Quah notes his work and that of others, including Kremer, Onatski and Stock, is consistent with the view that the global poor are many in number and likely to be so for a very long time. In addition, as Quah (1996c, 1997, 2001) and Bulli (2001) discuss, the process of discretizing the state space of a continuous variable is necessarily arbitrary and can alter the probabilistic properties of the data. Especially relevant here is the fact that the shape of the ergodic distribution can be altered by changing the discretization scheme. Reichlin (1999) demonstrates that the dynamic behavior inferred from the analysis of
31 As Quah (1993b, footnote 4) explains, the estimated ergodic distributions “. . . should not be read as
forecasts of what will happen in the future . . . ” (his emphasis). Rather, he continues, they “. . . should be interpreted simply as characterizations of tendencies in the post-War history that actually realized”.
Ch. 8: Growth Econometrics
597
Markov transition probabilities, and the apparent long-run implications of that behavior, are sensitive to the discretization scheme employed; this work also shows that the estimated ergodic distribution can be sensitive to small changes in the transition probabilities. Bulli (2001) addresses this critique and shows how to discretize the state space in a way that preserves the probabilistic properties of the data. Applying her method to cross-country income data she finds an estimated ergodic distribution quite different from that found by arbitrary discretization as well as being an accurate approximation to the distribution computed using a continuous state space method. An alternative formulation of distribution dynamics that avoids discretization problems is proposed by Quah (1996c, 1997) and models the cross-country income distribution at time t with the density function, ft (x). If the process describing the evolution of the distribution is again assumed to be time-invariant and first-order Markov, then ∞ density at time t + τ , τ > 0, will be ft+τ (x) = 0 gτ (x|z)ft (z) dz where gτ (x|z) is the τ -period-ahead density of x conditional on z. The function gτ (x|z) is the continuous analog of the transition matrix M and, assuming it exists, the ergodic ∞(long-run) density function, f∞ (x), implied by gτ (x|z) is the solution to f∞ (x) = 0 gτ (x|z)f∞ (z) dz. Using nonparametric methods, Quah (1996c, 1997) estimates various gτ (x|z) and finds strong evidence of twin-peakedness in the cross-country income distribution. The estimated ergodic densities presented by Bulli (2001) and Johnson (2004) support Quah’s conclusions. Azariadis and Stachurski (2003) derive the form of the gτ (x|z) implied by a stochastic version of the model in Azariadis and Drazen (1990). Estimation of the model’s parameters enables them to compute forward projections of the sequence of cross-country income distributions, and ultimately the ergodic distribution, implied by the model. Consistent with the work of Quah (1996c, 1997) they find bimodality to be a pervasive feature of the sequence of distributions for about 100 years. Eventually, however, all countries transition to the rich mode so the ergodic distribution is unimodal as found by Kremer, Onatski and Stock (2001). As Quah (2001) notes, there is “as yet” no theory of inference for this case so reconciliation of this result with the view that the ergodic distribution is bimodal cannot be done through formal statistical tests. However, while Quah (2001) observes that such a theory is an “obvious next step”, he suggests that we may be close to the limits of what can be reasonably inferred from the cross-country income data. Johnson (2000) offers an interpretation of gτ (x|z) which draws an analogy between the median of the conditional distribution and the law of motion of a non-stochastic one m(x) gτ (z|x) dz = variable dynamic system. The median is the function m(x) such that 0 0.5 so that a country with income of m(x) at time t has an equal chance of having a higher or lower income at time t + τ . Consider a point x0 such that m(x0 ) = x0 and suppose that, in some neighborhood of x0 , m(x) > x for x < x0 and m(x) < x for x > x0 implying Pr(xt+τ > xt ) > 0.5 for x < x0 and Pr(xt+τ < xt ) > 0.5 for x > x0 so that, in this neighborhood, countries with incomes different from x0 tend to move toward x0 . In the long run we may expect to find many countries in the vicinity of x0 creating the tendency for a mode in the ergodic density, f∞ (x), at x0 . Similarly,
598
S.N. Durlauf et al.
in a non-stochastic one-variable dynamic system with the law of motion xt+τ = m(xt ), the condition on the phase diagram for the local stability of a steady-state at x0 is that the graph of m(x) intersects the 45◦ line from above at x0 . In both cases, x0 is a point of accumulation in the sense that the long-run probability of finding countries in the vicinity of x0 will tend to be high relative to that elsewhere. Conversely, just as steady states are unstable in the non-stochastic case when m(x) crosses the 45◦ line from below, analogous points in the stochastic case tend to produce antimodes in the ergodic density. While Quah’s estimated gτ (x|z) indicate a strong tendency towards polarization in the world income distribution, they do not reveal much about intra-distribution mobility. Bimodality is arguably of less concern in a normative sense if there is movement between the two modes than it is if there is none. Quah (1996c) studies the mobility within the distribution by computing, (through stochastic simulation) the mean time for a “growth miracle” which he defines as passage from the 10th to 90th percentile of the distribution. He finds an expected time of 201 years for such a miracle to occur. Quah’s methods have subsequently been applied to a range of contexts. Andres and Lamo (1995) apply these methods to the OECD, Lamo (2000) to the regions of Spain, Johnson (2000) to US states, Bandyopadhyay (2002) to the Indian states, and Andrade et al. (2004) to Brazilian municipalities. These methods have also been extended to broader notions of distributional dynamics. Fiaschi and Lavezzi (2004) develop an analysis of the joint distribution of income levels and growth rates; their findings are compatible with the existence of multiple equilibria in the sense that countries may become trapped in the lower part of the income distribution. 4.3.4. Relationship between distributional convergence and the persistence of initial conditions Distributional methods have proved important in establishing stylized facts concerning the world income distribution. At the same time, there has been relatively little formal effort to explore the implications of findings such as twin peaks for the empirical salience of alternative growth theories. Some potential implications of distributional dynamics for evaluating theories are suggested by Quah (1996c), who finds that conditioning on measures of physical and human capital accumulation similar to those used by Mankiw, Romer and Weil (1992) and a dummy variable for the African continent has little effect on the dynamics of the cross-country income distribution. The polarization and immobility features are similar in both cases and conditioning increases the expected time for a growth miracle to 760 years.32 These results suggest that the heterogeneity revealed by the distributional approaches is, at least in part, due to the existence of convergence clubs. 32 Other efforts to find determinants of polarization and immobility have produced mixed results. For the
OECD countries, Andres and Lamo (1995) condition on the steady state implied by the Solow model and find little decrease in the tendency to polarization unless country specific effects are permitted. Lamo (2000) finds only a small increase in mobility for Spanish regions after conditioning on interregional migration flows. Bandyopadhyay (2002) shows that infrastructure spending and education measures appear to contribute to polarization between rich and poor states of India.
Ch. 8: Growth Econometrics
599
That being said, in general, it is relatively difficult to interpret properties of the crosscountry income distribution in the context of economic convergence in the sense of (22). To see why this is so, it is useful to focus on the absence of a clear relationship between β-convergence, which measures the relative growth of rich versus poor countries and σ -convergence, which focuses explicitly on the distribution of countries. These two convergence notions do not have any necessary implications for one another, i.e. one may hold when the other does not. For our purposes, what is important is that σ -convergence is not an implication of β-convergence and so does not speak directly to the question of the transience of contemporary income differences. The erroneous assertion that βconvergence implies σ -convergence is known as Galton’s fallacy and was introduced into the modern economic growth context by Friedman (1992) and Quah (1993a). To understand the fallacy, suppose that log per capita output in each of N countries obeys the AR(1) process log yi,t = α + ς log yi,t−1 + εi,t ,
(36)
where 0 < ς < 1 and the random variables εi,t are i.i.d across countries and time. For this model, each country will, by definition (22), exhibit convergence as any contemporaneous difference in output between two countries will disappear over time. Further, it is easy to see, using γi = T −1 (log yi,t+T − log yi,t ), that the regression of growth on a constant and initial income will exhibit β-convergence. This is immediate when one considers growth between t and t + 1 which means that growth obeys γi,t = α + (ς − 1) log yi,t−1 + εi,t ,
(37)
where ς − 1 < 0 by assumption. In this model, by construction, the unconditional population variance of log output is constant because the reduction in cross-section variance associated with the tendency of high-income countries to grow more slowly than low-income countries is offset by the presence of the random shocks εi,t . This indicates why σ -convergence is not a natural implication of long run independence from initial conditions; rather σ -convergence captures the evolution of the cross-section income distribution towards an invariant measure. This suggests that an important next step in the distributional approach to convergence is the development of tools which will allow distribution methods to more directly adjudicate substantive growth questions as they relate to the persistence of initial conditions. 4.4. Time series approaches to convergence A final approach to convergence is based on time series methods. This approach is largely statistical in nature, which allows various hypotheses about convergence to be precisely defined, and thereby reveals appropriate strategies for formal testing. A disadvantage of the approach is that it is not explicitly tied to particular growth theories. Bernard and Durlauf (1995, 1996), Evans (1998) and Hobijn and Franses (2000) provide a systematic framework for time series convergence tests.
600
S.N. Durlauf et al.
Following Bernard and Durlauf (1995), a set of countries I is said to exhibit convergence if lim Proj(log yi,t+T − log yj,t+T | Ft ) = 0 ∀i, j ∈ I,
T →∞
(38)
where Proj(a|b) denotes the projection of a on b and Ft denotes some information set; operationally, this information set will typically contain various functions of time and current and lagged values of log yi,t and log yj,t . Relative to our previous discussion, this definition represents a form of unconditional convergence that is closely related to (22). One can modify the definition to apply to the residual of per capita income after it has been projected on control variables such as savings rates in order to produce a definition of conditional convergence, but this has apparently not been done in the empirical literature. In evaluating (38), researchers have generally focused on whether deterministic or stochastic trends are present in log yi,t − log yj,t ; the presence of such trends immediately implies a violation of (38). As such, time series tests of convergence have typically been implemented using unit root tests. One reason for this focus is that the presence of unit roots in log yi,t − log yj,t allows for an extreme and therefore particularly interesting form of divergence between economies since a unit root implies that the difference log yi,t − log yj,t will, with probability 1, become arbitrarily large at some point in the future. The use of unit root and related time series tests has important implications for the sorts of countries that may be tested. Time series tests presuppose that yi,t may be thought of as generated by an invariant process in either levels or first differences, i.e., either levels or first differences may be modeled as the sum of deterministic terms plus a Wold representation for innovations. Such an assumption has significant economic content. As argued by Bernard and Durlauf (1996) countries that start far from their invariant distributions and are converging towards them, as occurs for countries that are in transition to the steady-state in the Solow–Swan model, will be associated with log yi,t − log yj,t series that do not fulfill this requirement. Hence, tests of (38) can produce erroneous results if applied to such economies. To see this intuitively, suppose that for country i, log yi,t = log yi,t+1 for all t, so that country has converged to a constant steady-state. Suppose that country j has the same steady-state as country i and is monotonically converging to this state so that log yi,t > log yj,t for all observations. Then log yi,t − log yj,t > 0 for all t in the sample; which means that the series has a nonzero mean and tests that fail to account for the fact that the density of log yi,t − log yj,t is changing across time can easily give erroneous inferences. For example one may use a test and conclude log yi,t − log yj,t possesses a nonzero mean and erroneously interpret this as evidence against convergence, when the fact that the process does not have a time-invariant mean is ignored. This argument suggests that time series convergence tests are really only appropriate for advanced economies that may plausibly be thought of as characterized by invariant distributions. Generally, the first generation of these tests rejected convergence for countries as well as other economic units. For example, Bernard and Durlauf (1995), studying 15
Ch. 8: Growth Econometrics
601
advanced industrialized economies between 1900 and 1989 based on data developed in Maddison (1982, 1989), find little evidence that convergence is occurring; Hobijn and Franses (2000) similarly find little evidence of convergence across 112 countries taken from the Penn World Table for the period 1960–1989. The findings of nonconvergence in output levels are echoed in recent work by Pesaran (2004a) who employs convergence definitions that explicitly focus on the probability of large deviations, i.e. Equation (25). He finds little evidence of output level convergence using either the Maddison or Penn World Table data. Relatively little explicit attention has been paid to the question of systematically identifying convergence clubs using time series methods. One exception is Hobijn and Franses (2000) who employ a clustering algorithm to identify groups of converging countries.33 Their algorithm finds many small clusters in their sample of 112 countries – depending on the particular rule used to determine cluster membership, they find 42 or 63 clusters with most containing just two or three countries. Hobijin and Frances view these clusters as convergence clubs but it is not clear that they represent groups of countries in distinct basins of attraction of the growth process. Absent controls for structural characteristics, these groupings could simply reflect the pattern of differences in those characteristics rather than differences in long-run outcomes due to differences in initial conditions. Moreover, the Bernard and Durlauf (1996) argument about the substantive economic assumptions that underlie time series methods for studying convergence seems applicable here. Given the breadth of the sample used by Hobijn and Franses, it is unlikely that it contains only data generated by countries whose behavior is near their respective steady-states; such an assumption is much more plausible for restricted samples such as the OECD countries. The clusters they find could thus reflect, in many cases at least, transition dynamics rather than convergence clubs. An important extension of this work would be the exploration of how one can distinguish convergence clubs from what may be called “transition” clubs, i.e. groups of countries exhibiting similar transition dynamics. A number of studies of time series convergence have criticized these claims of nonconvergence; these criticisms are based upon inferential issues that have arisen in the general unit roots literature. One of these issues concerns the validity of unit root tests in the presence of structural breaks in log yi,t − log yj,t ; as argued initially by Perron (1989), the failure to allow for structural breaks when testing for unit roots can lead to spurious evidence in support of the null hypothesis that a unit root is present. An initial analysis of this type in cross-country contexts is Greasley and Oxley (1997) who, imposing breaks exogenously, find convergence for Denmark and Sweden whereas the sort of test employed by Bernard and Durlauf (1995) does not. The role of breaks in 33 Corrado, Martin and Weeks (2004) extend this approach to allow for time variation in clusters. They con-
clude that there is substantial evidence of club convergence as opposed to overall convergence for European regions. A nice feature of their analysis is the effort to interpret the clubs that are identified statistically with alternative economic theories, and conclude that geographic proximity and demographic similarity correlate with their observed clusters.
602
S.N. Durlauf et al.
time series convergence tests is systematically studied in Li and Papell (1999). An important feature of their analysis is that Li and Papell avoid exogenous imposition of trend breaks and in fact find that the dates of these breaks exhibit some heterogeneity, although many of them cluster around World War II. Li and Papell find that the evidence for OECD convergence is more mixed than did Bernard and Durlauf (1995) in the sense that allowing for trend breaks reduces the number of country pairs that fail to exhibit convergence. Related findings are due to Carlino and Mills (1993) who study US regions and reject convergence except under specifications that allow for a trend break in 1946. These conclusions are shown by Loewy and Papell (1996) to hold even if one allows potential trend breaks to be endogenously determined by the data. While the analysis of trend breaks and convergence tests is valuable because of its implications about the time series structure of output differences between countries, studies of this type suffer from some interpretation problems. The presence of the regime break is presumably suggestive of an absence of convergence in the sense of (22) or (38), since it implies that there is some component of log yi,t − log yj,t that will not disappear over a sufficiently long time horizon. The time series definition of convergence is violated by any long-term predictability in output differences. Hence, claims by authors that allowing for data breaks produces evidence of convergence begs the question of what is meant by convergence. That being said, the sort of violation of (22) or (38) implied by a trend break is different from the type implied by a unit root. In particular, a break associated with the level of output means that the output difference between two countries is always bounded, unlike the unit root case. A distinct line of criticism of time series convergence tests is due to Michelacci and Zaffaroni (2000) who argue that convergence tests based on the presence of unit roots may perform badly when the true processes exhibit long memory. Let γ (L)ui,j,t denote the moving average representation for log yi,t −log yj,t . Suppose that the kth coefficient in the representation has the property that γk ∝ k d−1 ,
0 < d < 1.
(39)
In this case, shocks die out at a hyperbolic rather than geometric rate, which is one definition of long memory in a time series process. Michelacci and Zaffaroni (2000) show that if output deviations exhibit long memory, one can reconcile the claim of β-convergence with time series evidence of divergence, i.e., the failure of various tests to reject the presence of a unit root in per capita output deviations. This is a potentially important reconciliation of these two distinct testing strategies. That being said, the plausibility of a long memory characterization has yet to be established in the economics literature. One problem is that there is an absence of a body of economic theory that predicts the presence of long memory.34 The existing theoretical 34 There are at least two reasons why unit roots stem naturally from existing economic theories. First, tech-
nology shocks are generally modeled as permanent. Second, Euler equations often produce unit root or near unit root like conditions. The random walk theory of stock prices is one example of this, in which risk neutral agents produce unpredictability of stock price changes as an equilibrium.
Ch. 8: Growth Econometrics
603
justifications of long memory processes derive from aggregation arguments originating with Granger (1980); the conditions under which aggregation produces long memory do not have any particular empirical justification. In addition, there are questions concerning the ability of conventional statistical methods to allow one to distinguish between long memory models and various alternatives. Diebold and Inoue (2001) indicate how long memory may be spuriously inferred for series subject to regime shifts, so the strength of evidence of long memory cited by Michelacci and Zaffaroni (2000) may be questioned. Nevertheless, the Michelacci–Zaffaroni argument is important, not least because it focuses attention on the role in growth empirics of size and power issues that arise in all unit root contexts. Time series approaches to convergence are melded with analysis related to σ convergence in Evans (1996) who considers the cross-section variance of growth rates at time t, σt2 =
2 1 log yi,t − log y t , N i
(40)
where log y t = N1 i log yi,t and N is the cardinality of I . Evans observes that σt2 may be represented as a unit root process with a quadratic time trend when there is no cointegration among the series log yi,t . This leads Evans to suggest a time-series test of convergence based on unit root tests applied to σt2 . Employing this test, Evans concludes that there is convergence to a common trend among 13 industrial countries. One interpretation problem with this analysis is that it allows different countries to possess different deterministic trends in per capita output albeit with the same trend growth rate. Such differences are obviously germane with respect to convergence as an economic concept being consistent, for example, with the club and conditional convergence hypotheses but not with the unconditional convergence hypothesis. Evans (1997) provides a time series approach to estimating rates of convergence. He shows that OLS applied to Equation (18) yields a consistent estimator of β, and hence the rate of convergence, only if (i) each log yi,t − log y t obeys an AR(1) process having the same AR(1) parameter lying strictly between 0 and 1; and, (ii) the control variables, Xi and Zi , account for all cross-country heterogeneity. He argues that neither condition is likely to hold and offers an alternative method of measuring the rate of convergence based on the supposition that log yi,t − log y t follows an AR(q) process with lag polynomial Λ(L). Again, this specification allows countries to follow different parallel balanced growth paths and Evans defines the rate of convergence for economy i as the rate at which log yi,t “is expected to revert toward its balanced growth path far in the future”. He shows that, given that it is a real, distinct, positive fraction, the dominant root of the polynomial zq Λ(z−1 ) equals one minus this rate. Evans computes estimates of the convergence rates and their 90% confidence intervals for a sample of 48 countries over the period 1950–90 and for the contiguous US states over the period 1929–91. For the states, about a third of the point estimates are negative and about two-thirds of the confidence intervals contain zero, while for the countries, about half of the point estimates are negative and all but
604
S.N. Durlauf et al.
two of the confidence intervals contain zero. However, in spite of these positive estimated average convergence rates of 15.5% and 5.9% respectively, Evans’ analysis fails to yield persuasive evidence in favor of the conditional convergence hypothesis since, in most cases, the hypothesis of a convergence rate of zero cannot be rejected at the 10% level of significance. Later sections of the chapter will discuss how growth researchers can draw on time series data in other ways. One popular route has been to use panel data, with repeated observations on each country or region. Another method is to use techniques broadly similar to those of event studies in empirical finance, and trace out the consequences of specific events, such as major political or economic reforms. We will consider these approaches in Section 6.3 below. 4.5. Sources of convergence or divergence Abramovitz (1986), Baumol (1986), DeLong (1988) and many others, both before and since, view convergence as the process of follower countries “catching up” to leader countries by adopting their technologies. Some more recent contributors, such as Barro (1991) and Mankiw, Romer and Weil (1992), adopt the view that convergence is driven by diminishing returns to factors of production.35 In the neoclassical model, if each country has access to the same aggregate production function the steady-state is independent of an economy’s initial capital and labor stocks and hence initial income. In this model, long-run differences in output reflect differences in the determinants of accumulation, not differences in the technology used to combine inputs to produce output. Mankiw (1995, p. 301), for example, argues that for “understanding international experience, the best assumption may be that all countries have access to the same pool of knowledge, but differ by the degree to which they take advantage of this knowledge by investing in physical and human capital”. Even if one relaxes the assumption that countries have access to the same production function, convergence in growth rates can still occur so long as each country’s production function is concave in capital per efficiency unit of labor and each country experiences the same rate of labor-augmenting technical change. Klenow and Rodríguez-Clare (1997a) challenge this “neoclassical revival” with results suggesting that differences in factor accumulation are, at best, no more important
35 When an economy is below its steady-state value of capital per efficiency unit of labor, the marginal
product of capital is relatively high (and is higher than in the steady state). As a result, a given investment rate translates into relatively high output growth. Capital grows as well but, because of diminishing returns, the capital–output ratio rises and the marginal product of capital declines, causing the growth of output and capital to slow. Eventually, the economy converges to a steady state in which capital and output grow at the same rate and the marginal product of capital is sustained at a constant level by labor-augmenting technical progress. Dowrick and Rogers (2002) find that both diminishing returns and technology transfer are important contributors to the convergence process. See also Bernard and Jones (1996) and Barro and Sala-i-Martin (1997).
Ch. 8: Growth Econometrics
605
than differences in productivity in explaining the cross-country distribution of output per capita. They find that only about half of the cross-country variation in the 1985 level of output per worker is due to variation in human and physical capital inputs while a mere 10% or so of the variation in growth rates from 1960 to 1985 reflects differences in the growth of these inputs. The differences between the results of Mankiw, Romer and Weil (1992) and the findings of Klenow and Rodríguez-Clare (1997a) in their reexamination of Mankiw, Romer and Weil have two principal origins. First, citing concerns about the endogeneity of the input quantities, Klenow and Rodríguez-Clare (1997a) eschew estimation of the capital shares and choose to impute parameters based on the results of other studies. Second, they modify Mankiw, Romer and Weil’s measure of human capital accumulation by supplementing secondary school enrollment rates using data on primary enrollment. This yields a measure of human capital accumulation with less cross-country variation than that used by Mankiw, Romer and Weil. This one modification decreases the relative contribution of cross-country variation in human and physical capital inputs to variation in the 1985 level of output per worker to 40% from the 78% found by Mankiw, Romer and Weil. Prescott (1998) and Hall and Jones (1999) confirm the view that differences in inputs are unable to explain observed differences in output and Easterly and Levine (2001, p. 177) state that “[t]he ‘residual’ (total factor productivity, TFP) rather than factor accumulation accounts for most of the income and growth differences across countries”. Unlike many authors, who estimate TFP as a residual after assuming a common Cobb–Douglas production function, Henderson and Russell (2004) use a nonparametric production frontier approach (data envelopment analysis) to decompose the 1965 to 1990 growth of labor productivity into (i) shifts in the (common, worldwide) production frontier (technological change); (ii) movements toward (or away from) the frontier (technological catch-up); and (iii) capital accumulation. They find a dominant role for capital accumulation in the growth of the cross-country mean of labor productivity with human and physical capital each accounting for about half of that role.36 They also observe that the distribution of labor productivity became more dispersed from 1965 to 1990 and their results suggest that physical and human capital accumulation were largely responsible for the increased dispersion. The results of Henderson and Russell (2004) and those of the previous authors are, however, more consistent than it may seem. Klenow and Rodríguez-Clare (1997a), Hall and Jones (1999) and Barro and Sala-i-Martin (2004) argue that the standard growth accounting decomposition overstates the contribution of capital accumulation to output growth by attributing to capital the effect on output of increases in capital induced by increases in TFP. This effect also applies to Henderson and Russell’s approach and adjusting for it provides some reconciliation of their findings with those of Klenow 36 Note that any misspecification of the production function due to the Cobb–Douglas assumption in other
studies will tend to increase the apparent variation in TFP relative to that found by Henderson and Russell (2004) under the weaker assumption of constant returns to scale. In a rare effort to evaluate the Cobb–Douglas specification, Duffy and Papageorgiou (2000) reject it in favor of a more general CES functional form.
606
S.N. Durlauf et al.
and Rodríguez-Clare (1997a), Prescott (1998) and Hall and Jones (1999). The standard growth accounting formula attributes a fraction (equal to labor’s share of output) of the growth in output per worker to growth in TFP and a fraction (equal to capital’s share of output) to capital accumulation despite the fact that, in the steady-state, growth in output per worker is entirely due to technological progress [Barro and Sala-i-Martin (2004, pp. 457–460) and Klenow and Rodríguez-Clare (1997a, p. 75, fn. 4)]. The total effect of technological progress on output growth can thus be estimated by dividing labor’s share into the estimated growth rate of TFP. Interpreting “capital” broadly, labor’s share is about 1/3 suggesting that this effect is about three times the rate of growth of TFP. Henderson and Russell (2004, Table 5, row (a)) find that, on average, about 90% of the increase in output per worker over the 1965 to 1990 period is attributable to the accumulation of human and physical capital with increases in TFP accounting for the remaining 10%. Applying the adjustment discussed above suggests that technological progress accounts for about 30% of the growth in output per worker over this period while capital accumulation, due to transition dynamics, accounts for the remainder. As well as determining the relative contributions of inputs and TFP to the crosscountry variation in output and output growth, some have studied what features of the cross-country output distribution are explained by the cross-country distributions of inputs and TFP. Henderson and Russell (2004) document the emergence of a second mode in the cross-country distribution of output per worker between 1965 and 1990 and find changes in efficiency (the distance from the world technological frontier) to be largely responsible. A primary role for TFP in determining the shape of the long-run distribution of output per capita is found by Feyrer (2003) who uses Markov transition matrices estimated with data from 90 countries over the period 1970 to 1989 to estimate the ergodic distributions of output per capita, the capital–output ratio, human capital per worker, and TFP. He finds that the long-run distributions of both output per capita and TFP are bimodal while those of both the capital–output ratio and human capital per worker are unimodal. This result, Feyrer observes, has potentially important implications for theoretical modelling of development traps. It suggests that models of multiple equilibria that give rise to equilibrium differences in TFP are more promising than models that emphasize indeterminacy in capital intensity or educational attainment.37 It is also consistent with Quah’s (1996c) finding that conditioning on measures of physical and human capital accumulation (and a dummy variable for the African continent) has little effect on the dynamics of the cross-country income distribution. As discussed in Section 4.3.3, the shapes of ergodic distributions computed from transition matrices estimated with discretized data are not, in general, robust to changes in the way in which the state space is discretized. To avoid these problems, Johnson (2004) extends Feyrer’s analysis using Quah’s (1996c, 1997) continuous state-space methods and finds evidence of bimodality in the long-run distributions of both the capital–output
37 Romer (1993) discusses the intellectual origins of the centrality of capital accumulation in models of eco-
nomic development and argues that “idea gaps are central to the process of economic development” (p. 548).
Ch. 8: Growth Econometrics
607
ratio and TFP in addition to that in the long-run distribution of output per capita. This finding is broadly consistent with data produced by a version of the Solow growth model that includes a threshold externality à la Azariadis and Drazen (1990) but may be partly due to the computation of TFP after supposing a Cobb–Douglas production function across countries. Accordingly, some care must be exercised when drawing conclusions from these results. More generally, in much of the development accounting literature cited above, TFP is measured as a residual under the assumption of a concave worldwide production function. Durlauf and Johnson (1995) present evidence contrary to that assumption and in support of the implied multiple steady states in the growth process. It seems likely that the imposition of a concave production function in this case will tend to exaggerate the measured differences in TFP and so confound inferences about the importance of TFP variation.38 While Henderson and Russell (2004)’s approach is nonparametric and free from any assumption of a particular technology per se, it estimates the world technology frontier by fitting a convex cone to data on outputs and inputs. The imposed convexity of the production set prevents the method from discovering any nonconvexities that may exist and, in addition to masking the presence of multiple steady states, convexifying these nonconvexities would tend to overstate the cross-country variation in TFP. The extent to which our current understanding of the relative contributions of variation in inputs and variation in TFP to the observed variation in income levels is influenced by the effects on measured TFP of a misspecified worldwide technology remains an open research question. Despite these concerns and the differences in the precise estimates found by different researchers, it is clear that cross-country variation in inputs falls short of explaining the observed cross-country variation in output. The result that the TFP residual, a “measure of our ignorance” computed as the ratio of output to some index of inputs, is an important (perhaps the dominant) source of cross-country differences in long-run economic performance is useful but hardly satisfying and the need for a theory of TFP expressed by Prescott (1998) is well founded. Research such as Acemoglu and Zilibotti (2001) and Caselli and Coleman (2003) are promising contributions to that agenda. 5. Statistical models of the growth process While the convergence hypothesis plays a uniquely prominent role in empirical growth studies, it by no means represents the bulk of empirical growth studies. The primary focus of empirical growth papers may be thought of as a general exploration of potential growth determinants. This work may be divided into three main categories: (1) studies designed to establish that a given variable does or does not help explain cross-country growth differences, (2) efforts to uncover heterogeneity in growth, and (3) studies that 38 Graham and Temple (2003) show that the existence of multiple steady states can increase the variance and
accentuate bimodality in the observed cross-country distribution of TFP.
608
S.N. Durlauf et al.
attempt to uncover nonlinearities in the growth process. While analyses of these types are typically motivated by formal theories, operationally they represent efforts to develop statistical growth models that are consistent with certain types of specification tests. Section 5.1 discusses the analysis of how specific determinants affect growth. We describe the range of different variables that have appeared in growth regressions and consider alternative methodologies for analyzing growth models in the presence of uncertainty about which regressors should be included to define the “true” growth model. Section 5.2 addresses issues of parameter heterogeneity. The complexity of the growth process and the plethora of new growth theories suggest that the mapping of a given variable to growth is likely a function of both observed and unobserved factors; for example, the effect of human capital investment on growth may depend on the strength of property rights. We explore methods to account for parameter heterogeneity and consider the evidence that has been adduced in support of its presence. Section 5.3 focuses on the analysis of nonlinearities and multiple regimes in the growth process. Endogenous growth theories are often highly nonlinear and can produce multiple steady states in the growth process, both of which have important implications for econometric practice. This subsection explores alternative specifications that have been employed to allow for nonlinearity and multiple regimes and describes some of the main findings that have appeared to date. 5.1. Specifying explanatory variables in growth regressions In the search for a satisfactory statistical model of growth, the main area of effort has concerned the identification of appropriate variables to include in linear growth regressions, this generally amounts to the specification of Z in Equation (18). Appendix B provides a survey of different regressors that have been proposed in the growth literature with associated studies that either represent the first use of the variable or a well known use of the variable.39 The table contains 145 different regressors, the vast majority of which have been found to be statistically significant using conventional standards.40 One reason why so many alternative growth variables have been identified is due to questions of measurement. For example, a claim that domestic freedom affects growth leaves unanswered how freedom is to be measured. We have therefore organized the body of growth regressors into 43 distinct growth “theories” (by which we mean conceptually distinct growth determinants); each of these theories is found to be statistically significant in at least one study. As Appendix B indicates, the number of growth regressors that have been identified approaches the number of countries available in even the broadest samples. And 39 Our choices of which studies to include should not be taken to reflect any stance on any cases where there
is disagreement about priority as to who first proposed a variable. 40 Of course, the high percentage of statistically significant growth variables reflects publication bias as well
as data mining.
Ch. 8: Growth Econometrics
609
this regressor list does not consider cases where interactions between variables or nonlinear transformations of variables have been included as regressors; both of which are standard ways of introducing nonlinearities into a baseline growth regression. This plethora of potential regressors starkly illustrates one of the fundamental problems with empirical growth research, namely, the absence of any consensus on which growth determinants ought to be included in a growth model. In this section, we discuss efforts to address the question of variable choice in growth models. To make this discussion concrete, define Si as the set of regressors which a researcher always retains in a regression and let Ri denote additional controls in the regression, so that γi = ψSi + πRi + εi .
(41)
Notice that the inclusion of a variable in S does not mean the researcher is certain that it influences growth, only that it will be included in all models under consideration. To make this concrete, one can think of an exercise in which one wants to consider the relationship between initial income and growth. A researcher may choose to include initial income and the other Solow growth regressors in every specification of the model, but may in contrast be interested in the effects of different non-Solow growth regressors on inferences about the initial income/growth connection. If one takes the regressors that comprise R as fixed, then statements about elements of ψ are straightforward. A frequentist approach to inference will compute an estimate of the parameter ψˆ with an associated distribution that depends on the data generating process; Bayesian approaches will compute a posterior probability density of ψ given the researcher’s prior, the data, and the assumption that the linear model is correctly specified, i.e. the choice of variables in R corresponds to the “true” model. Designating the available data as D and a particular model as m, this posterior may be written as µ(ψ|D, m). The basic problem in developing statistical statements either about ψˆ or µ(ψ|D, m) is that there do not exist good theoretical reasons to specify a particular model m. This is not to say that the body of growth theories may not be used to identify candidates for R. Rather, the problem is that growth theories are, using a phrase due to Brock and Durlauf (2001a), open-ended. Theory open-endedness means that the growth theories are typically compatible with one another. For example, a theory that institutions matter for economic growth is not logically inconsistent with a theory that emphasizes the role of geography in growth. Hence, if one has a set of K potential growth theories, all of which are logically compatible with one another (and all subsets of theories), there exist 2K − 1 potential theoretical specifications of the form (41), each one of which corresponds to a particular combination of theories. One approach to resolving the problem of model uncertainty is based on identifying variables whose empirical importance is robust across different model specifications. This is the idea behind Levine and Renelt’s (1992) use of extreme bounds analysis [Leamer (1983) and Leamer and Leonard (1983)] to assess growth determinants. To see how extreme bounds analysis may be applied to the assessment of robustness of
610
S.N. Durlauf et al.
growth determinants, suppose that one has specified a space of possible models M. For model m, the growth process is γi = ψm Si + πm Ri,m + εi,m ,
(42)
where the subscripts m reflect the model specific nature of the parameters and associated residuals. One can compute ψˆ m for every model in M. Motivated by Leamer (1983), Levine and Renelt employ the rule that there is strong evidence that a given regressor in S, call it sl , robustly affects growth if the sign of the associated regression coefficient ψˆ l,m is constant and the coefficient estimate is statistically significant across all model specifications in M. In this analysis the S vector is composed of a variable of interest and other variables whose presence is held fixed across specifications. In the Levine and Renelt (1992) analysis, S includes the constant term, initial income, the investment share of GDP, secondary school enrollment rates, and population growth; these variables proxy for those suggested by the Solow model. Models are distinguished by alternative combinations of 1 to 3 variables taken from a set of 7 variables; these correspond to alternative choices of Ri,m . Based on the constant sign and statistical significance criteria, Levine and Renelt (1992) conclude that the only robust growth determinants among the elements of Si are initial income and the share of investment in GDP. These two findings are confirmed in subsequent work by Kalaitzidakis, Mamuneas and Stengos (2000) who allow for potential nonlinearities in (41). Specifically, they consider partially linear versions of (41), so that γi = ψm Si + fm (πRi,m ) + εi,m .
(43)
Note that the function f (·) is allowed to vary across specifications of R. As in Levine and Renelt (1992), Kalaitzidakis, Mamuneas and Stengos conclude that initial income and physical capital investment rates are robust determinants of growth. Unlike Levine and Renelt, they also find that inflation volatility and exchange rate distortions are robust; this is interesting as it is an example where the failure to account for nonlinearity in one set of variables masks the importance of another. From a decision-theoretic perspective, the extreme bounds approach is a problematic methodology. The basic difficulty, discussed in detail in Brock and Durlauf (2001a) and Brock, Durlauf and West (2003) is that if one is interested in ψl because one is considering whether to change si,l , by one unit, i.e. one is advising country i on a policy change, the extreme bounds standard corresponds to a very risk averse way of responding to model uncertainty. Specifically, suppose that for a policymaker, El(si,l , m) represents the expected loss associated with the current policy level in country i. We assume that one is only interested in the case where an increase in the policy raises growth, which means we will assume that it is necessary for ψˆ l,m > 0 in order to conclude that one should make the change. One can approximate the t-statistic rule, i.e., requiring that the coefficient estimate for sl be statistically significant in order to justify a policy as implying that El(si,l + 1, m) − El(si,l , m) = ψˆ l,m − 2sd(ψˆ l,m ) > 0, (44)
Ch. 8: Growth Econometrics
611
where sd(ψˆ l,m ) is the estimate of the standard deviation associated with ψˆ l,m and the statistical significance level required for the coefficient is assumed to correspond to a t-statistic of 2. This loss function may look odd, but it is in fact the sort of loss function implicitly assumed whenever one relies on t-statistics to make policy decisions. Extreme bounds analysis requires that (44) holds for every model in M. This requires that El(si,l ), the expected loss for a policymaker when one conditions only on the policy variable, has the property that El(si,l + 1) − El(si,l ) > 0 ⇒ El(si,l + 1, m) − El(si,l , m) > 0
∀m.
(45)
This means that the policymaker must have minimax preferences with respect to model uncertainty, i.e. he will make the policy change only if it yields a positive expected payoff under the least favorable model in the model space. While there are reasons to believe that in practice, individuals assess model uncertainty differently than withinmodel uncertainty,41 the extreme risk aversion embedded in (45) seems hard to justify. Even when one moves away from decision-theoretic considerations, extreme bounds analysis is somewhat difficult to interpret as a statistical procedure. Hoover and Perez (2004), for example, show that the use of extreme bounds analysis can lead to the conclusion that many growth determinants are fragile even when they are part of the data generating process. They also find that the procedure has poor power properties in the sense that some regressors that do not matter may spuriously appear to be robust.42 The concern that extreme bounds analysis represents an excessively conservative approach to evaluating empirical results led Sala-i-Martin (1997a, 1997b) to propose a different way to evaluate the robustness of findings. Within a model, suppose there is an evaluative criterion for ψˆ m that is used to determine whether the variable sl matters for the growth process. One example of such a standard is whether or not ψˆ l,m is statistically significant at some level. Sala-i-Martin first proposes averaging the statistical significance levels via Sˆi = (46) ωˆ m Sˆi,m , m
ˆ where Sˆi,m is the statistical significance level associated with ψm and ωˆ m is the weight assigned to model m, m ωˆ m = 1. Sala-i-Martin (1997a, 1997b) employs weights determined by the likelihoods of each model as well as employing equal weighting. Second, Sala-i-Martin (1997a, 1997b) proposes examining the percentage of times a variable appears statistically significant with a given sign; a variable whose sign and statistical significance holds across 95% of the different models estimated is regarded as robust. This approach finds that initial income, the investment to GDP ratio and secondary school education are all robust determinants of growth. Sala-i-Martin (1997a,
41 See discussion in Brock, Durlauf and West (2003) of the Ellsberg Paradox. 42 For further discussion of extreme bounds analysis, see Temple (2000b) and the references therein.
612
S.N. Durlauf et al.
1997b) extends this analysis to the evaluation of additional variables and finds a number also are robust by his criteria. While these approaches have the important advantage over extreme bounds analysis of accounting for the informational content of the entire distribution of ψˆ m , the procedures do not have any decision-theoretic or conventional statistical justification. We are unaware of any statistical interpretation to averaged significance levels. Further, little is understood about the statistical properties of these procedures. Hoover and Perez (2004), for example, find that the second Sala-i-Martin procedure has poor size properties, in the sense that “true” growth determinants are still likely to fail to be identified. Dissatisfaction with extreme bounds analysis and the variants we have described have led some authors to embed the determinants of robust growth regressors in a general model selection context. Hendry and Krolzig (2004) and Hoover and Perez (2004) both employ general-to-specific modeling methodologies generally associated with the research program of David Hendry [cf. Hendry (1995)] to select one version of (41) out of the model space. In both papers, the linear model that is selected out of the space of possible models is one where growth is determined by years an economy is open, the rate of equipment investment, a measure of political instability based on the number of coups and revolutions, a measure of the percentage of the population that is Confucian and a measure of the percentage of the population that is Protestant. Methodologically, these papers in essence employ algorithms which choose a particular regression model from a space of models through comparisons based on a set of statistical tests. The extent to which one finds this approach appealing is a function of the extent to which one is sympathetic to the general methodological foundations of the Hendry research program; we avoid such an extended evaluation here, but simply note that like other general prescriptions the program remains controversial, especially the extent to which it relies on automatic model selection procedures that do not possess a clear decision-theoretic justification. As such, it is somewhat unclear how to evaluate the output of the procedure in terms of the objectives of a researcher. That being said, the automated procedures Hendry works with have the important virtue that they can facilitate identifying small sets of models that are well supported by available data. Identification of such models is important, for example, in forecasting, where Hendry’s procedures appear to have a strong track record. In our judgment, the most promising current approach to accounting for model uncertainty employs model averaging techniques to construct parameter estimates that formally address the dependence of model-specific estimates on a given model. Examples where model averaging has been applied to cross-country growth data include Brock and Durlauf (2001a), Brock, Durlauf and West (2003), Doppelhofer, Miller and Sala-i-Martin (2004), Fernandez, Ley and Steel (2001a) and Masanjala and Papageorgiou (2004). The basic idea in this work is to treat the “true” growth model43 as an 43 In this discussion, we will assume that one of the models in the model space M is the correct specification
of the growth process. When none of the model specifications is the correct one, this naturally affects the interpretation of the model averaging procedure.
Ch. 8: Growth Econometrics
613
unobservable variable. In order to account for this variable, each element m in the model space M is associated with a posterior model probability µ(m|D). By Bayes’ rule, µ(m|D) ∝ µ(D|m)µ(m),
(47)
where µ(D|m) is the likelihood of the data given the model and µ(m) is the prior model probability. These model probabilities are used to eliminate the dependence of parameter analysis on a specific model. For frequentist estimates, averaging is done across the model-specific estimates ψˆ m to produce an estimate ψˆ via ψˆ = (48) ψˆ m µ(m|D) m
whereas for the Bayesian context, the dependence of the posterior probability measure of the parameter of interest, µ(ψ|D, m) on the model choice is eliminated via standard conditional probability arguments, i.e. µ(ψ|D) = (49) µ(ψ|D, m)µ(m|D). m∈M
Brock, Durlauf and West (2003) argue that the strategy of constructing posterior probabilities that are not model-dependent is the appropriate one when the objective of the statistical exercise is to evaluate alternative policy questions such as whether to change elements of Si by one unit. Notice that this approach assumes that the goal of the exercise is to study a parameter, i.e. ψ, not to identify the best growth model. Model averaging approaches are still quite new in the growth literature, so many questions exist as to how to implement the procedure. One issue concerns the specification of priors on parameters within a model. Brock and Durlauf (2001a), Brock, Durlauf and West (2003), and Doppelhofer, Miller and Sala-i-Martin (2004) assume a diffuse prior on the model specific coefficients. The advantage of this prior is that, when the errors are normal with known variance, the posterior expected value of ψ, conditional on the data D and model m, is the ordinary least squares estimator ψˆ m . The disadvantage of this approach is that since the diffuse prior on the regression parameters is improper, one has to be careful that the posterior model probabilities associated with the prior are interpretable. For this reason, Doppelhofer, Miller and Sala-i-Martin (2004) eschew reference to their methodology as strictly Bayesian. That being said, so long as the posterior model probabilities include appropriate penalties for model complexity [and Brock and Durlauf (2001a), Brock, Durlauf and West (2003), and Doppelhofer, Miller and Sala-i-Martin (2004) all compute posterior model probabilities using BIC adjusted likelihoods] we do not see any conceptual problem in interpreting this approach as strictly Bayesian. Fernandez, Ley and Steel (2001a) and Masanjala and Papageorgiou (2004) employ proper priors and therefore avoid such concerns.44 We are unaware
44 Fernandez, Ley and Steel (2001b) provide a general analysis of proper model specific priors for model
averaging exercises.
614
S.N. Durlauf et al.
of any evidence that the choice of prior for the within-model regression coefficients is of great importance in terms of empirical inferences for the growth contexts that have been studied; Masanjala and Papageorgiou (2004) in fact compare results using proper priors with the improper priors we have described and find that the choice of prior is unimportant. A second unresolved issue concerns the specification of the prior model probabilities µ(m). In the model averaging literature, the general assumption has been to assign equal prior probabilities to all models in M. This prior may be interpreted as assuming that the prior probability that a given variable appears in the “true” model is 0.5 and that the probability that one variable appears in the model is independent of whether others appear. Doppelhofer, Miller and Sala-i-Martin (2004) consider modifications of this prior in which the probability that a given variable appears in the true model is p < 0.5; these alternative probabilities are chosen in order to assign greater weight to more parsimonious growth models, i.e. models in which fewer regressors appear.45 Brock and Durlauf (2001a) and Brock, Durlauf and West (2003) argue against the assumption that the probability that one regressor should appear in a growth model is independent of the inclusion of others. The basic problem with priors that assume independence is analogous to the red bus/blue bus problem in discrete choice theory; namely, some regressors are quite similar to others, e.g., alternative measures of trade openness, whereas other regressors are quite disparate, e.g., geography and institutions. Brock, Durlauf and West (2003) propose a tree structure to organize model uncertainty for linear growth models. First, they argue that growth models suffer from theory uncertainty. Hence, one can identify alternative classes of models based on what growth theories are included. Second, for each specification of a body of theories to be embedded, they argue there is specification uncertainty. A given set of theories requires determining whether the theories interact, whether they are subject to threshold effects or other types of nonlinearity, etc. Third, for each theory and model specification, there is measurement uncertainty. The statement that weather affects growth does not specify the relevant empirical proxies, e.g., the number of sunny days, average temperature, etc. Finally, each choice of theory, specification and measurement is argued to suffer from heterogeneity uncertainty, which means that it is unclear which subsets of countries obey a common linear model. Brock, Durlauf and West (2003) argue that one should assign priors that account for the interdependences implied by this structure in assigning model probabilities. Appendix B follows this approach in organizing growth regressors according to theory. Doppelhofer, Miller and Sala-i-Martin (2004) and Fernandez, Ley and Steel (2001a) employ model averaging methods to identify which growth regressors should be included in linear growth models. These analyses do not distinguish between variables to
45 In our judgment, this presumption is unappealing as our own prior beliefs suggest that the true growth
model is likely to contain many distinct factors. One implication of the open-endedness of growth theories is that the simultaneous importance of many factors is certainly plausible.
Ch. 8: Growth Econometrics
615
be included in all regressions and variables whose inclusion determines alternative models; all variables are pooled and all possible combinations are considered. Doppelhofer, Miller and Sala-i-Martin (2004) working with 31 potential growth determinants, conclude, weighting prior models so that the expected number of included regressors is 7 (this corresponds to a prior probability of variable inclusion of about 0.25), that four variables have posterior model inclusion probabilities above 0.9: initial income, fraction of GDP in mining, number of years the economy has been open,46 and fraction of the population following Confucianism. Working with a universe of 41 potential growth determinants, Fernandez, Ley and Steel find that, under the assumption that the prior probability that a given variable appears in the correct growth model is 0.5, four variables have posterior model inclusion probabilities above 0.9: initial income, fraction of the population following Confucianism, life expectancy, and rate of equipment investment. Brock and Durlauf (2001a) and Masanjala and Papageorgiou (2004) employ model averaging to study the reason for the poor growth performance of sub-Saharan Africa. Brock and Durlauf (2001a) reexamine Easterly and Levine’s (1997a) finding that ethnic heterogeneity helps explain sub-Saharan Africa’s growth problems. This reanalysis finds that the Easterly and Levine (1997a) claim is robust in the sense that ethnic heterogeneity helps explain why growth in sub-Saharan Africa had stagnated relative to the rest of the world. On the other hand, Brock and Durlauf (2001a) also find that ethnic heterogeneity does not appear to explain growth patterns in the rest of the world. This leads to the unresolved question of why ethnic heterogeneity has uniquely strong growth effects in sub-Sahran Africa. Masanjala and Papageorgiou (2004) conduct a general analysis of the determinants of sub-Saharan African growth versus the world as a whole and conclude that the relevant growth variables for Africa are quite different. In particular, variation in sub-Saharan growth is much more closely associated with the share of the economy made up by primary commodities production. They also find, contrary to Doppelhofer, Miller and Sala-i-Martin (2004) that the share of mining in the economy is a robust determinant of growth in Africa but not the world as a whole. Finally, model averaging has been applied by Brock, Durlauf and West (2003) to analyze the question of how to employ growth regressions to evaluate policy recommendations. Specifically, the paper assesses the question of whether a policymaker should favor a reduction of tariffs for sub-Saharan African countries; the analysis assumes that the policymaker possesses mean/variance preferences with respect to the effects of changes in current policies with a constant tradeoff of mean against standard deviation of 1 to 2. The analysis finds strong support for a tariff reduction in that it concludes that a policymaker with these preferences should support a tariff reduction for any of the countries in sub-Saharan Africa unless the policymaker has a very strong prior that sub-Saharan African countries obey a distinct linear growth process from the rest of the world. In the case where the policymaker has a strong prior that sub-Saharan
46 Sachs and Warner (1995) use this variable as an index of overall openness of an economy.
616
S.N. Durlauf et al.
Africa is “different” from the rest of the world, there is sufficient uncertainty about the relationship between tariffs and growth for these countries that a change in the rates cannot be justified; the strong prior in essence means that the growth experiences of non-African countries have little effect on the precision of estimates of growth behavior that are constructed using data on sub-Saharan African countries in isolation. 5.2. Parameter heterogeneity From its earliest stages, the use of linear growth models has generated considerable unease with respect to the statistical foundations of the exercise. Arguably, the data for very different countries cannot be seen as realizations associated with a common data generating process (DGP). For econometricians that have been trained to search for a good approximation to a DGP, the modeling assumptions and procedures of the growth literature can look arbitrary. One expression of this concern is captured in a famous remark in Harberger (1987): “What do Thailand, the Dominican Republic, Zimbabwe, Greece, and Bolivia have in common that merits their being put in the same regression analysis?” Views differ on the extent to which this objection is fundamental. There is general agreement that, when studying growth, it will be difficult to recover a DGP even if one exists. In particular, the prospects for recovering causal effects are clearly weak. Those who are only satisfied with the specification and estimation of a structural model, in which parameters are either ‘deep’ or correspond to precisely defined causal effects within a coherent theoretical framework, will be permanently disappointed.47 The growth literature must have a less ambitious goal, namely to investigate whether or not particular hypotheses have any support in the data. In practice, growth researchers are looking for patterns and systematic tendencies that can increase our understanding of the growth process, in combination with historical analysis, case studies, and relevant theoretical models. Another key aim of empirical growth research, which is harder than it looks at first sight, is to communicate the degree of support for any patterns identified by the researcher. The issue of parameter heterogeneity is essentially that raised by Harberger. Why should we expect disparate countries to lie on a common surface? Clearly this criticism could be applied to most empirical work in social science, whether the data points reflect the actions and characteristics of individuals and firms, or the aggregations of their choices that are used in macroeconometrics. What is distinctive about the growth context is not so much the lack of a common surface, as the way in which the sample size limits the scope for addressing the problem. In principle, one response would be to choose a more flexible model that has a stronger chance of being a good approximation to the process generating the data. Yet this can be hard, and an inherently fragile procedure, when the sample is rarely greater than 100 observations. 47 Note that this reflects the shortcomings of economic theory as well as those of data and econometric
analysis.
Ch. 8: Growth Econometrics
617
If parameter heterogeneity is present, the consequences are potentially serious, except in a special case. If a slope parameter varies randomly across units, and is distributed independently of the variables in the regression and the disturbances, the coefficient estimate should be an unbiased estimate of the mean of the parameter. The assumption of independence is not one, however, that may be expected in light of the body of growth theories. For example, when estimating the relationship between growth and investment, the marginal effect of investment will almost certainly be correlated with aspects of the economic environment that should also be included in the regression. The solution to this general problem is to change the specification in a way that allows greater flexibility in estimation. There are many ways of doing this. One approach is to consider more general functional forms than the canonical Solow regression which for comparison purposes we restate as: γi = k + β log yi,0 + πn log(ni + g + δ) + πK log sK,i + πH log sH,i + εi . (50) Liu and Stengos (1999) estimate a semiparametric partially linear version of this model, namely γi = k + fβ (log yi,0 ) + πn log(ni + g + δ) + πK log sK,i + fπH (log sH,i ) + εi ,
(51)
where fβ (·) and fπH (·) are arbitrary (except for variance smoothness requirements) functions. One important finding is that the marginal effect of initial income is only negative when initial per capita income exceeds about $1800. They also find a threshold effect in secondary school enrollment rates (their empirical proxy for log sH,i ) so the variable is only associated with a positive impact on growth if it exceeds about 15%. Banerjee and Duflo (2003) use this same regression strategy to study nonlinearity in the relationship between changes in inequality and growth; their specification estimates a version of (51) where initial income and human capital investment enter linearly (along with some additional non-Solow variables) but with the addition on the right-hand side of the function fG (Gi,t − Gi,t−5 ) where Gi,t is the Gini coefficient. Using a panel of 45 countries and 5-year growth averages, their analysis produces an estimate of fG (·) which has an inverted U shape. One limitation of such studies is that they only allow for nonlinearity for a subset of growth determinants, an assumption that has little theoretical justification and is, from a statistical perspective, ad hoc; of course the approach is more general and less ad hoc then simply assuming linearity as is done in most of the literature. Durlauf, Kourtellos and Minkin (2001) extend this approach and estimate a version of the augmented Solow model that allows the parameters for each country to vary as functions of initial income, i.e. γi = k(yi,0 ) + β(yi,0 ) log yi,0 + πn (yi,0 ) log(ni + δ + g) + πK (yi,0 ) log sK,i + πH (yi,0 ) log sH,i + εi .
(52)
This formulation means that each initial income level defines a distinct Solow regression; as such it shifts the focus away from nonlinearity towards parameter heterogeneity,
618
S.N. Durlauf et al.
although the model is of course nonlinear in yi,0 . This approach reveals considerable parameter heterogeneity especially among the poorer countries. Durlauf, Kourtellos and Minkin (2001) confirm Liu and Stengos (1999) in finding that β(yi,0 ) is positive for low yi,0 values and negative for higher ones. They also find that πK (yi,0 ) fluctuates greatly over the range of yi,0 values in their sample. This work is extended in Kourtellos (2003a) who finds parameter dependence on initial literacy and initial life expectancy. The varying coefficient approach is also employed in Mamuneas, Savvides and Stengos (2004) who analyze annual measures of total factor productivity for 51 countries. They consider a regression model of TFP in which the coefficient on human capital in the regression is allowed to depend on human capital both in isolation and in conjunction with a measure of trade openness (other coefficients are held constant). Constancy of the human capital coefficient is rejected across a range of specifications. At a minimum, it generally makes sense for empirical researchers to test for neglected parameter heterogeneity, either using interaction terms or by carrying out diagnostic tests. Chesher (1984) showed that White’s information matrix test can be used in this context. For the normal linear model with fixed regressors, Hall (1987) showed that, asymptotically, the information matrix test corresponds to a joint test for heteroskedasticity and non-normality. Later in the chapter, we discuss how evidence of heteroskedasticity should sometimes be seen as an indicator of misspecification. Other authors have attempted to employ panel data to identify parameter heterogeneity without the imposition of a functional relationship between parameters and various observable variables. An important early effort is Canova and Marcet (1995). Defining si,t as the logarithm of the ratio of a country’s per capita income to the time t international aggregate value, Canova and Marcet estimate models of the form si,t = ai + ρi si,t−1 + εi,t .
(53)
The long-run forecast of si,t is given by ai /(1 − ρi ) with 1 − ρi being the rate of convergence towards that value. Canova and Marcet estimate their model using data on the regions of Europe and on 17 Western European countries. Restricting the parameters ai and ρi to be constant across i gives a standard β-convergence test and yields an estimated annual rate of convergence of approximately 2%. On the other hand, allowing for heterogeneity in these parameters produces a “substantial”, statistically significant, dispersion of the implied long-run si,t forecasts. Moreover, those forecasts are positively correlated with si,0 , the initial values of si,t , implying a dependence of long-run outcomes on initial conditions contrary to the convergence hypothesis. For the countrylevel data, differences in initial conditions explain almost half the cross-sectional variation in long-run forecasts; in contrast, the role of standard control variables such as rates of physical and human capital accumulation and government spending shares is minor. The latter finding must be tempered by the fact that the sample variation in these controls is less than that in Barro (1991) or Mankiw, Romer and Weil (1992), for example. A similar approach is taken by Maddala and Wu (2000) who consider models of the form log yi,t = αi + ρi log yi,t−1 + ui,t ,
(54)
Ch. 8: Growth Econometrics
619
which is of course very similar to the model analyzed by Canova and Marcet (1995). Employing shrinkage estimators for αi and ρi , they conclude that convergence rates, measured as βi = − log ρi exhibit substantial heterogeneity. 5.3. Nonlinearity and multiple regimes In this section we discuss several papers that have attempted to disentangle the roles of heterogeneous structural characteristics and initial conditions in determining growth performance. These studies employ a wide variety of statistical methods in attempting to identify how initial conditions affect growth. Despite this, there is substantial congruence in the conclusions of these papers as these studies each provide evidence of the existence of convergence clubs even after accounting for variation in structural characteristics. An early contribution to this literature is Durlauf and Johnson (1995) who use classification and regression tree (CART) methods to search for nonlinearities in the growth process as implied by the existence of convergence clubs.48 The CART procedure identifies subgroups of countries that obey a common linear growth model based on the Solow variables. These subgroups are identified by initial income and literacy, a typical subgroup l is defined by countries whose initial income lies within the interval ϑ l,y yi,0 < ϑ¯ l,y and whose literacy rate Li lies in the interval ϑ l,L Li < ϑ¯ l,L . The number of subgroups and the boundaries for the variable intervals that define them are chosen by an algorithm that trades off model complexity (i.e. the number of subgroups) and goodness of fit. Because the procedure sequentially splits the data into finer and finer subgroups, it gives the data a tree structure. Durlauf and Johnson (1995) also test the null hypothesis of a common growth regime against the alternative hypothesis of a growth process with multiple regimes in which economies with similar initial conditions tend to converge to one another. Using income per capita and the literacy rate (as a proxy for human capital) to measure the initial level of development and, using the same cross-country data set as Mankiw, Romer and Weil, Durlauf and Johnson reject the single regime model required for global convergence. That is, even after controlling for the structural heterogeneity implied by Mankiw, Romer and Weil’s augmented version of the Solow model, there is a role for initial conditions in explaining variation in cross-country growth behavior. Durlauf and Johnson’s (1995) findings of multiple convergence clubs appear to be reinforced by subsequent research. Papageorgiou and Masanjala (2004) note that one
48 A detailed discussion of regression tree methods appears in Breiman et al. (1984). The technical appendix
of Durlauf and Johnson (1995) presents a treatment tailored to the specific question of identifying multiple regimes in growth models. Regression tree methods suffer from the absence of a well-developed asymptotic theory for testing the number of regimes that are present in a data set, but the procedure is consistent in the sense that under relatively weak conditions, if there are a finite number of regimes, as the sample size grows to infinity, the correct model will be revealed.
620
S.N. Durlauf et al.
possible source for Durlauf and Johnson’s findings may occur due to the misspecification of the aggregate production function. As observed in Section 2, the linear representation of the Solow model represents an approximation around the steady-state when the aggregate production function is Cobb–Douglas. Papageorgiou and Masanjala estimate a version of the Solow model based on a constant elasticity of substitution (CES) production function rather than the Cobb–Douglas, following findings in Duffy and Papageorgiou (2000). They then examine the question of whether or not Durlauf and Johnson’s multiple regimes remain under the CES specification. Using Hansen’s (2000) approach to sample splitting and threshold estimation, they find statistically significant evidence of thresholds in the data. The sample splits they estimate divide the data in four distinct growth regimes and are broadly consistent with those found by Durlauf and Johnson.49 These findings are extended in recent work due to Tan (2004) who employs a procedure known as GUIDE (generalized, unbiased interaction detection and estimation) to identify subgroups of countries which obey a common growth model.50 Relative to CART, the GUIDE algorithm has two advantages: (1) the algorithm explicitly looks for interactions between explanatory variables when identifying splits, and (2) some within model testing supplements the penalties for model complexity and thereby reduces the tendency for CART procedures to produce an excessive number of splits in finite samples. Tan (2004) finds strong evidence that measures of institutional quality and ethnic fractionalization define convergence clubs across a wide range of countries. He also finds weaker evidence that geography distinguishes the growth process for sub-Saharan Africa from the rest of the world. Further research has corroborated the evidence of multiple regimes using alternative statistical methods. One approach that has proved useful is based on projection pursuit methods.51 Desdoigts (1999) uses these methods in an attempt to separate the roles of microeconomic heterogeneity and initial conditions in the growth experiences of a group of countries and identifies groups of countries with relatively homogeneous growth experiences based on data about the characteristics and initial conditions of each country. The idea of projection pursuit is to find the orthogonal projections of the data into low-dimensional spaces that best display some interesting feature of the data. A well-known special case of projection pursuit is principal components analysis. 49 Motivated by the debate over trade openness and growth, Papageorgiou (2002) applies Hansen’s method
to the Durlauf and Johnson data with the trade share added to the set of variables on which sample splits may occur. He finds that this variable divides the middle-income countries into high and low growth groups obeying different growth processes; however openness does not appear to matter for high and low income countries. This suggests the importance of further work on which variables are most appropriate in characterizing threshold effects. Using the regression tree approach with a large collection of candidate split variables, Johnson and Takeyama (2001) find evidence of thresholds in US state economic growth behavior defined by variables likely to be proxies for the capital/labor ratio, agglomeration effects, and communication effects. 50 GUIDE originates in Loh (2002). 51 Projection pursuit is developed in Friedman and Tukey (1974) and Friedman (1987). Appendix A of Desdoigts (1999) provides a useful primer.
Ch. 8: Growth Econometrics
621
In principal components analysis, one takes only as many components as are necessary to account for “most” of the variation in the data. Similarly, in projection pursuit one should only consider as many dimensions as needed to account for “most” of the clustering in the data. Desdoigts finds several interesting clusters. The first is the OECD countries. The two projections identifying this cluster put most of their weight on the primary and secondary school enrollment rates, the 1960 income gap and the rate of growth in the labor force. The prominence of variables that Desdoigts argues are proxies for initial conditions among those defining the projections leads him to conclude that initial conditions are more important in defining this cluster than are other country characteristics. Reapplication of the clustering method to the remaining (non-OECD) countries yields three sub-clusters that can be described as Africa, Southeast Asia, and Latin America. Here the projections put most weight on government consumption, the secondary school enrollment rate and investment in electrical machinery and transportation equipment. Most of these variables are argued to proxy for structural characteristics of the economies, suggesting that they, rather than initial conditions, are responsible for the differences in growth experiences across the three geographic sub-clusters. Nevertheless, this approach relies on the judgment of the researcher in determining which variables proxy for initial conditions and which proxy for structural characteristics. Further evidence of the utility of projection pursuit methods may be found in Kourtellos (2003b). Unlike Desdoigts, Kourtellos (2003b) uses projection pursuit to construct models of the growth process. Formally, he estimates models of the form γi =
L
fl (yi,0 βl + Xi ψl + Zi πl ) + εi .
(55)
l=1
Each element in the summation represents a distinct projection. Kourtellos uncovers evidence of two steady-states, one for low initial income and low initial human capital countries. A third approach to multiple regimes is employed by Bloom, Canning and Sevilla (2003) based on the observation that if long-run outcomes are determined by fundamental forces alone, the relationship between exogenous variables and income levels ought to be unique. If initial conditions play a role there will be multiple relationships – one for each basin of attraction defined by initial conditions. If there are two (stochastic) steady states, and large shocks are sufficiently infrequent,52 the system will, under suitable regularity conditions, exhibit an invariant probability measure that can be described by a “reduced form” model in which the long-run behavior of log yi,t depends only on the exogenous variables, mi , such as log yi,t = log y1∗ (mi ) + u1,i,t
with probability p(mi )
(56)
52 The assumed rarity of large shocks implies that movements between basins of attraction of each of the
steady states are sufficiently infrequent that they can be ignored in estimation. This assumption is consistent with, for example, Bianchi’s (1997) finding of very little mobility in the cross-country income distribution.
622
S.N. Durlauf et al.
and log yi,t = log y2∗ (mi ) + u2,i,t
with probability 1 − p(mi ),
(57)
where u1,i,t and u2,i,t are independent, zero-mean deviations from the steady-state log means log y1∗ (mi ) and log y2∗ (mi ) respectively, and p(mi ) is the probability that country i is in the basin of attraction of the first of the two steady states. From the perspective of the econometrician, log yi,t thus obeys a mixture process. The two steady states associated with (56) and (57) are possibly interpretable as a low-income regime or poverty trap and as a high-income or perpetual growth regime respectively. Bloom, Canning and Sevilla estimate a linear version of this model using 1985 income data from 152 countries with the absolute value of the latitude of the (approximate) center of each country as the fundamental exogenous variable. They are able to reject the null hypothesis of a single regime model in favor of the alternative of a model with two regimes – a high-level (manufacturing and services) steady state in which income is independent of latitude and a low-level (agricultural) steady-state in which income depends on latitude (presumably through its influence on climate). In addition, the probability of being in the high-level steady state is found to rise with latitude. A final approach to multiple regimes is due to Canova (2004) who introduces a procedure for panel data that estimates the number of groups and the assignment of countries or regions to these groups, drawing on Bayesian ideas. This approach has the important feature that it allows for parameter heterogeneity across countries within a given subgroup. The researcher can order the countries or regions by various criteria (for example, output per capita in the pre-sample period) and the estimation procedure then chooses break points and group membership in such a way that the predictive ability of the overall model is maximized. This approach is applied to autoregressive models of per capita output as in Equation (54) above. Using data on per capita income data in the regions of Europe, Canova (2004) finds that ordering the data by initial income maximizes the marginal likelihood of the model and breaks the data into 4 clusters. The estimated mean steady-states for each group are significantly different from each other implying that the groups are convergence clubs. The differences in the means are also economically important with the lowest and highest being 45% and 115% of the overall average respectively. Canova finds little across-group mobility especially among those regions that are initially poor. Using data on per capita income in the OECD countries, two clusters are found and, again, initial per capita income is the preferred ordering variable. The estimated model parameters imply an “economically large” long-run difference in the average incomes of countries in the two groups with little mobility between them. In assessing these analyses, it is important to recognize an identification problem in attempting to link evidence of multiple growth regimes to particular theoretical growth models. As argued in Durlauf and Johnson (1995), this identification problem relates to whether evidence of multiple regimes represents evidence of multiple steady-states as opposed to nonlinearity in the growth process.
Ch. 8: Growth Econometrics
623
Figure 8. Nonlinearity versus multiple steady-states.
To see why this is so, suppose that one has identified two sets of countries that obey separate growth regimes with regime membership determined by a country’s initial capital stock, i.e. there exists a capital threshold k T that divides the two groups of countries. An example of this can be seen in Figure 8. Clearly, the two sets of countries do not obey a common linear model but it is not clear whether or not multiple steady-states exist. The output behavior of low capital countries is compatible with either the solid or dashed curve in the lower part of the figure, but only the solid curve produces multiple steady-states. The identification problem stems from the fact that one does not have observations that allow one to distinguish differences in the long-run behavior of countries that start with capital stocks in the vicinity of k T . This argument does not depend on growth regimes determined by the capital stock but it does depend on whether or not the variable or variables that define the regimes are growing over time, as would occur for initial income or initial literacy. For growing variables, the possibility exists that countries currently associated with low levels of the variables will in the future exhibit behaviors that are similar to those countries which are currently associated with high levels of the variables. How might evidence of multiple steady-states be established? One possibility is via the use of structural models in empirical analysis. While this has not been done econometrically, Graham and Temple (2003) follow this strategy and calibrate a two-sector general equilibrium model with increasing returns to scale in nonagricultural production. Their empirically motivated choice of calibration parameters produces a model
624
S.N. Durlauf et al.
which implies that some countries are in a low-output equilibrium. Another possibility is to exploit time series variation in a single country to identify the presence of jumps from one equilibrium or steady state to another.
6. Econometric issues I: Alternative data structures Our discussion of growth econometrics now shifts from general issues of hypothesis testing and model specification to explore specific econometric issues that arise in the estimation of growth models. This section reviews econometric issues that arise for the different types of data structures that appear in growth analyses. By data structures, we refer to features such as whether the data are observed in cross-section, time series, or panel as well as to whether particular data series are conceptualized as endogenous or exogenous. At the risk of stating the obvious, choices of method involve significant trade-offs, which depend partly on statistical considerations and partly on the economic context. This means that attempts at universal prescriptions are misguided, and we will try to show the desirability of matching techniques to the economic question at hand. One example, to be discussed further below, would be the choice between panel data methods and the estimation of separate time series regressions for each country. The use of panel data is likely to increase efficiency and allow richer models to be estimated, but at the expense of potentially serious biases if the parameter homogeneity assumptions are incorrect. This trade-off between robustness and efficiency is another running theme of our survey. The scientific solution would be to base the choice of estimation method on a context-specific loss function, but this is clearly a difficult task, and in practice more subjective decisions are involved. This section has four main elements. Section 6.1 examines econometric issues that arise in the use of time series data to study growth, emphasizing some of the drawbacks of this approach. Section 6.2 discusses the many issues that arise when panel data are employed, an increasingly popular approach to growth questions. We consider the estimation of dynamic models in the presence of fixed effects, and alternatives to standard procedures. Section 6.3 describes another increasingly popular approach, namely the use of “event studies” to analyze growth behavior, based on studying responses to major shocks such as policy reforms. Section 6.4 examines endogeneity and the use of instrumental variables. We argue that the use of instrumental variables in growth contexts is more problematic than is often appreciated and suggest the importance of combining instrumental variable choice with a systematic approach to model selection. 6.1. Time series approaches At first glance, the most natural way to understand growth would be to examine time series data for each country in isolation. As we saw previously, growth varies substantially over time, and countries experience distinct events that contribute to this variation, such as changes in government and in economic policy.
Ch. 8: Growth Econometrics
625
In practice a time series approach runs into substantial difficulties. One key constraint is the available data. For many developing countries, some of the most important data are only available on an annual basis, with limited coverage before the 1960s. Moreover, the listing of annual data in widely used sources and online databases can be misleading, because some key variables are measured less frequently. For example, population figures are often based primarily on census data, while measures of average educational attainment are often constructed by interpolating between census observations using school enrollments. When examining published data, it is not always clear where this kind of interpolation has been used. The true extent of information in the time series variation may be less than appears at first glance, and conventional standard errors on parameter estimates will be misleading when interpolated data are used. Even where reliable data are available, some key growth determinants display relatively little time variation, a point that has been emphasized by Easterly et al. (1993), Easterly (2001) and Pritchett (2000a). There do exist other variables that appear to show significant variation, but this variation may not correspond to the concept the researcher has in mind. An example would be political stability. Since Barro (1991), researchers have sometimes used the incidence of political revolutions and coups as a measure of political instability. The interpretation of such an index clearly varies depending on the length of the time period used to construct it. If the hypothesis of interest relates to underlying political uncertainty (say, the ex-ante probability of a transfer of power) then the observations on political instability would need to be averaged over a long time period. The variation in political instability at shorter horizons only casts light on a different hypothesis, namely the direct impact of revolutions and coups. There are other significant problems with the time series approach. The hypotheses of most interest to growth theorists are mainly about the evolution of potential output, not deviations from potential output such as business cycles and output collapses. Since measured output is a noisy indicator of potential output, it is easy for the econometric modeling of a growth process to be contaminated by business cycle dynamics. A simple way to illustrate this would be to consider what happens if measured log output is equal to the log of potential output plus a random error. If log output is trend stationary, this is a classical measurement error problem. When lags of output or the growth rate are used as explanatory variables, the parameter estimates will be inconsistent. Such problems are likely to be even more serious in developing countries, where large slumps or crises are not uncommon, and output may deviate for long periods from any previous structural trend [Pritchett (2000a)]. We have already seen the extent to which output behaves very differently in developing countries compared to OECD members, and a major collapse in output is not a rare event. There may be no underlying trend in the sense commonly understood, and conventional time series methods should be applied with caution. Some techniques that are widely used in the literature on business cycles in developed countries, such as the Hodrick–Prescott filter, will often be inappropriate in the context of developing countries. The problem of short-run output instability extends further. It is easy to construct examples where the difference between observed output and potential output is corre-
626
S.N. Durlauf et al.
lated with variables that move up and down at high frequencies, with inflation being one obvious example. This means that time series studies of inflation and growth based on observed output will find it hard to isolate reliably an effect of inflation on potential output; for further discussion see Temple (2000a). When considerations like these are combined with the paucity of the available data, it appears a hard task to learn about long-term growth using time series regressions, especially when developing countries are the main focus of interest. Nevertheless, despite these problems, there are some hypotheses for which time series variation can be informative. We have already seen the gains from time series approaches to convergence issues. Jones (1995) and Kocherlakota and Yi (1997) show how time series models can be used to discriminate between different growth theories. To take the simplest example, the AK model of growth predicts that the growth rate will be a function of the share of investment in GDP. Jones points out that investment rates have trended upwards in many OECD countries, with no corresponding increase in growth rates. Although this might be explained by offsetting changes in other growth determinants, it does provide evidence against simple versions of the AK model. Jones (1995) and Kocherlakota and Yi (1997) develop a statistical test of endogenous growth models based on regressing growth on lagged growth and a lagged policy variable (or the lagged investment rate, as in Jones). Exogenous growth models predict that the coefficients on the lagged policy variable should sum to zero, indicating no long-run effect of permanent changes in this variable on the growth rate. In contrast, some endogenous growth models imply that the sum of coefficients should be non-zero. A simple time series regression then provides a direct test of the predictions of these models. More formally, as in Jones (1995), for a given country i one can investigate a dynamic relationship for the growth rate γi,t where γi,t = A(L)γi,t−1 + B(L)zi,t + εi,t ,
(58)
where z is the policy variable or growth determinant of interest, and A(L) and B(L) are lag polynomials assumed to be compatible with stationarity. The hypothesis of interest is whether B(1) = 0. If the sum of the coefficients in the lag polynomial B(L) is significantly different from zero, this implies that a permanent change in the variable z will affect the growth rate indefinitely. As Jones (1995) explicitly discusses, this test is best seen as indicating whether a policy change affects growth over a long horizon, rather than firmly identifying or rejecting the presence of a long-run growth effect in the theoretical sense of that term. The theoretical conditions under which policy variables affect the long-run growth rate are remarkably strict, and many endogenous growth models are best seen as new theories of potentially sizeable level effects.53 This approach is closely related to Granger-causality testing, where the hypothesis of interest would be the explanatory power of lags of zi,t for γi,t conditional on lagged 53 See Temple (2003) for more discussion of this point and the long-run implications of different growth
models.
Ch. 8: Growth Econometrics
627
values of γi,t . Blomstrom, Lipsey and Zejan (1996) carry out Granger-causality tests for investment and growth using panel data with five-year subperiods. They find strong evidence that lagged growth rates have explanatory power for investment rates, but much weaker evidence for causality in the more conventional direction from investment to growth. Hence, the partial correlation between growth and investment found in many cross-section studies may not reflect a causal effect of investment. In a similar vein, Campos and Nugent (2002) find that, once Granger-causality tests are applied, the evidence that political instability affects growth may be weaker than usually believed. The motivation for these two studies, and others like them, is that evidence of temporal precedence helps to build a case that one variable is influenced by another. When this idea is extended to panels, an underlying assumption is that timing patterns and effects will be similar across units (countries or regions). Potential heterogeneity has sometimes been acknowledged, as in the observation of Campos and Nugent (2002) that their results are heavily influenced by the African countries in the sample. The potential importance of these factors is also established in Binder and Brock (2004) who, by using panel methods to allow for heterogeneity in country-specific dynamics, find feedbacks from investment to growth beyond those that appear in Blomstrom, Lipsey and Zejan (1996). A second issue is more technical. Since testing for Granger-causality using panel data requires a dynamic model, the use of a standard fixed effects (within groups) estimator is likely to be inappropriate when individual effects are present. We discuss this further in Section 6.2 below. One potential solution is the use of instrumental variable procedures, as in Campos and Nugent (2002). In the context of investment and growth, a comprehensive examination of the associated econometric issues has been carried out by Bond, Leblebicioglu and Schiantarelli (2004). Their work shows that these issues are more than technicalities: unlike Blomstrom, Lipsey and Zejan (1996), they find strong evidence that investment has a causal effect on growth. A familiar objection to the more ambitious interpretations of Granger-causality is that much economic behavior is forward-looking [see, for example, Klenow and RodriguezClare (1997b)]. The movements of stock markets are one instance where temporal sequences can be misleading about causality. Similarly, when entrepreneurs or governments invest heavily in infrastructure projects, or when unusually high inflows of foreign direct investment are observed, the fact that such investments precede strong growth does not establish a causal effect. 6.2. Panel data As we emphasized above, the prospects for reliable generalizations in empirical growth research are often constrained by the limited number of countries available. This constraint makes parameter estimates imprecise, and also limits the extent to which researchers can apply more sophisticated methods, such as semiparametric estimators. A natural response to this constraint is to use the within-country variation to multiply the number of observations. Using different episodes within the same country is ulti-
628
S.N. Durlauf et al.
mately the only practical substitute for somehow increasing the number of countries. To the extent that important variables change over time, this appears the most promising way to sidestep many of the problems that face growth researchers. Moreover, as the years pass and more data become available, the prospects for informative work of this kind can only improve. We first discuss the implementation and advantages of panel data estimators in more detail, and then some of the technical issues that arise in the context of growth. Perhaps not surprisingly, these methods introduce a set of problems of their own, and should not be regarded as a panacea. Too often, panel data results are interpreted without sufficient care and risk leading researchers astray. In particular, we highlight the care needed in interpreting estimates based on fixed effects. We will use T to denote the number of time series observations in a panel of N countries or regions. At first sight, T should be relatively high in this context, because of the availability of annual data. But the concerns about time series analysis raised above continue to apply. Important variables are either measured at infrequent intervals, or show little year-to-year variation that can be used to identify their effects. Moreover, variation in growth rates at annual frequencies may give very misleading answers about the longer-term growth process. For this reason, most panel data studies in the growth field have averaged data over five or ten year periods. Given the lack of data before 1960, this implies that growth panels not only have relatively few cross-sectional units (the number of countries employed is often between 50 and 100) but also very low values of T , often 5 or 6 at most.54 Most empirical growth models estimated using panel data are based on the hypothesis of conditional convergence, namely that countries converge to parallel equilibrium growth paths, the levels of which are a function of a few variables. A corollary is that an equation for growth (essentially the first difference of log output) should contain some dynamics in lagged output. In this case, the growth equation can be rewritten as a dynamic panel data model in which current output is regressed on controls and lagged output, as in Islam (1995). In statistical terms this is the same model, the only difference of interpretation being that the coefficient on initial output (originally β) is now 1 + β: log yi,t = (1 + β) log yi,t−1 + ψXi,t + πZi,t + αi + µt + εi,t .
(59)
This regression is a general panel analog to the cross-section regression (18). In this formulation, αi is a country-specific effect and µt is a time-specific effect. The inclusion of time-specific effects is important in the growth context, not least because the means of the log output series will typically increase over time, given productivity growth at the world level. Inclusion of a country-specific effect allows permanent differences in the level of income between countries that are not captured by Xi,t or Zi,t . In principle, one can 54 This is true of the many published studies that have used version 5.6 of the Penn World Table. Now that
more recent data are available, there is more scope for estimating panels with a longer time dimension.
Ch. 8: Growth Econometrics
629
also allow the parameters 1 + β, ψ, and π to differ across i; Lee, Pesaran and Smith (1997, 1998) do this for the coefficients for log yi,t−1 and a linear time trend (the latter allowing for steady-state differences in the rate of technological change, corresponding to non-parallel growth paths in the steady state). The vast majority of panel data growth studies use a fixed effects (within-group) estimator rather than a random effects estimator. Standard random effects estimators require that the individual effects αi are distributed independently of the explanatory variables, and this requirement is clearly violated for a dynamic panel such as (59) by construction, given the dependence of log yi,t−1 on αi . Given the popularity of fixed effects estimators, it is important to understand how these estimators work. In a fixed effects regression there is a full set of country-specific intercepts, one for each country, and inference proceeds conditional on the particular countries observed (a natural choice in this context). Identification of the slope parameters, usually constrained to be the same across countries, relies on variation over time within each country. The “between” variation, namely the variation across countries in the long-run averages of the variables, is not used. The key strength of this method, familiar from the microeconometric literature, is the ability to address one form of unobserved heterogeneity: any omitted variables that are constant over time will not bias the estimates, even if the omitted variables are correlated with the explanatory variables. Intuitively, the country-specific intercepts can be seen as picking up the combined effects of all such variables. This is the usual motivation for using fixed effects in the growth context, especially in estimating conditional convergence regressions, as is further discussed in Islam (1995), Caselli, Esquivel and Lefort (1996) and Temple (1999). A particular motivation for the use of fixed effects arises from the Mankiw, Romer and Weil (1992) implementation of the Solow model. As discussion in Section 3, their version of the model implies that one determinant of the level of the steady-state growth path is the initial level of efficiency (Ai,0 ) and cross-section heterogeneity in it should usually be regarded as unobservable, cf. Equation (15). Islam (1995) explicitly develops a specification in which this term is treated as a fixed effect, while world growth and common shocks are incorporated using time-specific effects. The use of panel data methods to address unobserved heterogeneity can bring substantial gains in robustness, but is not without costs. The fixed-effects identification strategy cannot be applied in all contexts. Sometimes a variable of interest is measured at only one point in time. Even where variables are measured at more frequent intervals, some are highly persistent, in which case the within-country variation is unlikely to be informative. At one extreme, some explanatory variables of interest are essentially fixed factors, like geographic characteristics or ethnolinguistic diversity. Here the only available variation is “between-country”, and empirical work will have to be based on cross-sections or pooled cross-section time-series. Alternatively a two-stage hybrid of these methods can be used, in which a panel data estimator is used to obtain estimates of the fixed effects, which are then explicitly modeled in a second stage as in Hoeffler (2002). As we discuss further below, an important direction for future panel data work may be the analysis of the information content of country-specific effects.
630
S.N. Durlauf et al.
A common failing of panel data studies based on within-country variation is that researchers do not pay enough attention to the dynamics of adjustment. There are many panel data papers on human capital and growth that test only whether a change in school enrollment or years of schooling has an immediate effect on aggregate productivity, which seems an implausible hypothesis. Another example, given by Pritchett (2000a), is the use of panels to study inequality and growth. All too often, changes in the distribution of income are implicitly expected to have an immediate impact on growth. Yet many of the relevant theoretical papers highlight long-run effects, and there is a strong presumption that much of the short-run variation in measures of inequality is due to measurement error. In these circumstances, it is hard to see how the available within-country variation can shed much useful light. There is also a more general problem. Since the fixed effects estimator ignores the between-country variation, the reduction in bias typically comes at the expense of higher standard errors. Another reason for imprecision is that either of the devices used to eliminate the country-specific intercepts – the within-groups transformation or firstdifferencing – will tend to exacerbate the effect of measurement error.55 As a result, it is common for researchers using panel data models with fixed effects, especially in the context of small T , to obtain imprecise sets of parameter estimates. Given the potentially unattractive trade-off between robustness and efficiency, Barro (1997), Temple (1999), Pritchett (2000a) and Wacziarg (2002) all argue that the use of fixed effects in empirical growth models has to be approached with care. The price of eliminating the misleading component of the between variation – namely, the variation due to unobserved heterogeneity – is that all the between variation is lost. There are alternative ways to reveal this point, but consider the random effects GLS estimator of the slope parameters, which will be more efficient than the within-country estimator for small T when the random-effects assumptions are appropriate. This GLS estimator can be written as a matrix-weighted average of the within-country estimator and the between-country estimator, which is based on averaging the data over time and then estimating a simple cross-section regression by OLS.56 The weights on the two sets of parameter estimates are the inverses of their respective variances. The corollary of high standard errors using within-country estimation, indicating that the within-country variation is relatively uninformative, is that random effects estimates based on a panel of five-yearly averages are very similar to OLS estimates based on thirty-year averages [Wacziarg (2002)]. Informally, the random effects estimator sees the between-country variation as offering the greatest scope for identifying the parameters.57 55 See Arellano (2003, pp. 47–51) for a more formal treatment of this issue. 56 This result holds for the GLS estimator of the random effects model. In practice, since the true variance
components are unknown, feasible GLS must be used. 57 Of course, this does not imply that the random effects estimator is the best choice; as we have seen, the
underlying assumptions for consistency of the estimator are necessarily invalid for a dynamic panel. Instead, our discussion is intended to draw attention to the trade-off between bias and efficiency in deciding whether or not to use fixed-effects estimation.
Ch. 8: Growth Econometrics
631
This should not be surprising: growth episodes within countries inevitably look a great deal more alike than growth episodes across countries, and therefore offer less identifying variation. Restricting the analysis to the within variation eliminates one source of bias, but immediately makes it harder to identify growth effects with any degree of precision. This general problem is discussed in Pritchett (2000a). Many of the explanatory variables currently used in growth research are either highly stable over time, or tending to trend in one direction. Educational attainment is an obvious example. Without useful identifying variation in the time series data, the within-country approach is in trouble. Moreover, growth is quite volatile at short horizons. It will typically be hard to explain this variation using predictors that show little variation over time, or that are measured with substantial errors. The result has been a number of panel data studies suggesting that a given variable “does not matter” when a more accurate interpretation is that its effect cannot be identified using the data at hand. Some of these problems suggest a natural alternative to the within-country estimator, which is to devote more attention to modeling the heterogeneity, rather than treating it as unobserved [Temple (1999)]. To put this differently, current panel data methods treat the individual effects as nuisance parameters. As argued by Durlauf and Quah (1999) this is clearly inappropriate in the growth context. The individual effects are of fundamental interest to growth economists because they appear to be a key source of persistent income differences. This suggests that more attention should be given to modeling the heterogeneity rather than finding ways to eliminate its effects.58 Depending on the sources of heterogeneity, even simple recommendations, such as including a complete set of regional dummies, can help to alleviate the biases associated with omitted variables. More than a decade of growth research has identified a host of fixed factors that could be used to substitute for country-specific intercepts. A growth model that includes these variables can still exploit the panel structure of the data, and overall this approach has clear advantages in both statistical and economic terms. It means that the between variation is retained, rather than entirely thrown away, while the explicit modeling of the country-specific effects is directly informative about the sources of persistent income and growth differences. In practice, the literature has focused on another aspect of using panel data estimators to investigate growth. Nickell (1981) showed that within-groups estimates of a dynamic panel data model can be badly biased for small T , even as N goes to infinity. The direction of this bias is such that, in a growth model, output appears less persistent than it should (the estimate of β is too low) and the rate of conditional convergence will be overestimated. In other areas of economics, it has become increasingly common to avoid the withingroups estimator when estimating dynamic models. The most widely-used alternative 58 Note that fixed-effects estimators could retain a useful role, because it would be natural to compare their
parameter estimates with those obtained using a specific model for the heterogeneity. Where the estimates of common parameters, such as the coefficient on the lagged dependent variable, are different across the two methods, this could indicate the chosen model for the heterogeneity is misspecified.
632
S.N. Durlauf et al.
strategy is to difference the model to eliminate the fixed effects, and then use two stage least squares or GMM to address the correlation between the differenced lagged dependent variable and the induced MA(1) error term. To see the need for instrumental variable procedures, first-difference (59) to obtain log yi,t = (1 + β) log yi,t−1 + Xi,t ψ + Zi,t π + µt + εi,t − εi,t−1 (60) and note that (absent an unlikely error structure) the log yi,t−1 component of log yi,t−1 will be correlated with the εi,t−1 component of the new composite error term, as is clearly seen by considering Equation (59) lagged one period. Hence, at least one of the explanatory variables in the first-differenced equation will be correlated with the disturbances, and instrumental variable procedures are required. Arellano and Bond (1991), building on work by Holtz-Eakin, Newey and Rosen (1988), developed the GMM approach to dynamic panels in detail, including methods suitable for unbalanced panels and specification tests. Caselli, Esquivel and Lefort (1996) applied their estimator in the growth context and, as discussed above, this approach yielded a much faster rate of conditional convergence than found in cross-section studies. The GMM approach is typically based on using lagged levels of the series as instruments for lagged first differences. If the error terms in the levels equation (εit ) are serially uncorrelated then log yi,t−1 can be instrumented using log yi,t−2 and earlier lagged levels (where available). This corresponds to a set of moment conditions that can be used to estimate the first-differenced equation by GMM. Bond (2002) provides an accessible introduction to this approach. As an empirical strategy for growth research, this has some appeal, because it could alleviate biases due to measurement error and endogenous explanatory variables. In practice, many researchers are skeptical that lags are suitable instruments. It is easy to see that a variable such as educational attainment may influence output with a considerable delay, so that the exclusion of lags from the growth equation can look arbitrary. More generally, the GMM approach relies on a lack of serial correlation in the error terms of the growth equation (before differencing). Although this assumption can be tested using the methods developed in Arellano and Bond (1991), and can also be relaxed by an appropriate choice of instruments, it is nevertheless restrictive in some contexts. Another concern is that the explanatory variables may be highly persistent, as is clearly true of output. Lagged levels can then be weak instruments for first differences, and the GMM panel data estimator is likely to be severely biased in short panels. Bond, Hoeffler and Temple (2001) illustrate this point by comparing the Caselli, Esquivel and Lefort (1996) estimates of the coefficient on lagged output with OLS and within-group estimates. Since the OLS and within-group estimates of β are biased in opposing directions then, leaving aside sampling variability and small-sample considerations, a consistent parameter estimate should lie between these two extremes [see Nerlove (1999, 2000)]. Formally, when the explanatory variables other than lagged out-
Ch. 8: Growth Econometrics
633
put are strictly exogenous, we have p lim βˆW G < p lim βˆ < p lim βˆOLS
(61)
where βˆ is a consistent parameter estimate, βˆW G is the within-groups estimate and βˆOLS is the estimate from a straightforward pooled OLS regression. For the data set and model used by Caselli, Esquivel and Lefort, this large-sample prediction is not valid, which raises a question mark over the reliability of the first-differenced GMM estimates. One device that can be informative in short panels is to make more restrictive assumptions about the initial conditions. If the observations at the start of the sample are distributed in a way that is representative of steady-state behavior, in a sense that can be made more precise, efficiency gains are possible. Assumptions about the initial conditions can be used to derive a “system” GMM estimator, of the form developed and studied by Arellano and Bover (1995) and Blundell and Bond (1998), and also discussed in Ahn and Schmidt (1995) and Hahn (1999). In this estimator, not only are lagged levels used as instruments for first differences, but lagged first differences are used as instruments for levels, which corresponds to an extra set of moment conditions. There is some Monte Carlo evidence [Blundell and Bond (1998)] that this estimator is more robust than the Arellano–Bond method in the presence of highly persistent series. As also shown by Blundell and Bond (1998), the necessary assumptions can be seen in terms of an extra restriction, namely that the deviations of the initial values of log yi,t from their long-run values are not systematically related to the individual effects.59 For simplicity, we focus on the case where there are no explanatory variables other than lagged output. The required assumption on the initial conditions is that, for all i = 1, . . . , N we have
E (log yi,1 − y¯i )αi = 0, (62) where the y¯i are the long-run values of the log yi,t series and are therefore functions of the individual effects αi and the autoregressive parameter β. This assumption on the initial conditions ensures that E[ log yi,2 αi ] = 0
(63)
and this together with the mild assumption that the changes in the errors are uncorrelated with the individual effects, i.e. E[εi,t αi ] = 0 implies T − 2 extra moment conditions of the form
E log yi,t−1 (αi + εi,t ) = 0 for i = 1, . . . , N and t = 3, 4, . . . , T .
(64)
(65)
59 Note that the long-run values of log output are evolving over time when time-specific effects are included
in the model.
634
S.N. Durlauf et al.
Intuitively, as is clear from the new moment conditions, the extra assumptions ensure that the lagged first difference of the dependent variable is a valid instrument for untransformed equations in levels since it is uncorrelated with the composite error term in the levels equation. These extra moment conditions can then be combined with the more conventional conditions used in the Arellano–Bond method. This builds in some insurance against weak identification, because if the series are persistent and lagged levels are weak instruments for first differences, it may still be the case that lagged first differences have some explanatory power for levels.60 In principle, the validity of the restrictions on the initial conditions can be tested using the incremental Sargan statistic (or C statistic) associated with the additional moment conditions. Yet the validity of the restriction should arguably be evaluated in wider terms, based on some knowledge of the historical forces giving rise to the observed initial conditions. This point – that key statistical assumptions should not always be evaluated only in statistical terms – is one that we will return to later. Alternatives to GMM have been proposed. Kiviet (1995, 1999) derives an analytical approximation to the Nickell bias that can be used to construct a bias-adjusted withincountry estimator for dynamic panels. The simulation evidence reported in Judson and Owen (1999) and Bun and Kiviet (2001) suggests that this estimator performs well relative to standard alternatives when N and T are small. One minor limitation is that it cannot yet be applied to an unbalanced panel. A more serious limitation, relative to GMM, is that it does not address the possible correlation between the explanatory variables and the disturbances due to simultaneity and measurement error. Nevertheless, for researchers determined to use fixed effects estimation, there is a clear case for implementing this bias adjustment, at least as a complement to other methods. A further issue that arises when estimating dynamic panel data models is that of parameter heterogeneity. If a slope parameter such as β varies across countries, and the explanatory variable is serially correlated, this will induce serial correlation in the error term. If we focus on a simple case where a researcher wrongly assumes βi = β for all i = 1, . . . , N then the error process for a given country will contain a component that resembles (βi − β) log yi,t−1 . Hence there is serial correlation in the errors, given the persistence of output. The estimates of a dynamic panel data model will be inconsistent even if GMM methods are applied. This problem was analyzed in more general terms by Robertson and Symons (1992) and Pesaran and Smith (1995) and has been explored in great depth for the growth context by Lee, Pesaran and Smith (1997, 1998). Since an absence of serial correlation in the disturbances is usually a critical assumption for the GMM approach, parameter heterogeneity can be a serious concern. Some of the possible solutions, such as regressions applied to single time series, or the pooled mean group estimator developed by Pesaran, Shin and Smith (1999), have limitations in studying growth for reasons already discussed. An alternative solution is to split the sample into groups that are more likely to 60 An alternative approach would be to use small-sample bias adjustments for GMM panel data estimators,
such as those described in Hahn, Hausman and Kuersteiner (2001).
Ch. 8: Growth Econometrics
635
share similar parameter values. Groupings by regional location or level of development are a natural starting point. Perhaps the state of the art in analyzing growth using panel data and allowing for parameter heterogeneity is represented by Phillips and Sul (2003). They allow for heterogeneity in parameters not only across countries, but also over time. Temporal heterogeneity is rarely investigated in panel studies, but may be important, especially if observed growth patterns combine transitional dynamics towards a country’s steady state with fluctuations around that steady-state. Phillips and Sul find some evidence of convergence towards steady states for OECD economies as well as US regions. We close our discussion of panel data approaches by noting some unresolved issues in their application. It is important to be aware how panel data methods change the substantive interpretation of regression results, and care is needed when moving between the general forms of the estimators and the economic hypotheses under study. Relevant examples occur in analyses of β-convergence. If one finds β-convergence in a panel study having allowed for fixed effects, the interpretation of this finding is very different than if one finds evidence of convergence in the absence of fixed effects. Specifically, the presence of fixed effects represents an immediate violation of our convergence definitions (20) or (22) as different economies must exhibit steady-state differences in per capita income regardless of whether they have identical saving rates and population growth rates.61 Fixed effects may even control for the presence of unmodelled determinants of steady state growth, an identification problem analogous to the one that was previously discussed in the context of interpreting the control variables Z in Equations (17) and (18) above. Similarly, allowing for differences in time trends for per capita output, as done in Lee, Pesaran and Smith (1997, 1998) means that the finding of extremely rapid β-convergence is consistent with long-run divergence of per capita output across the economies they study; the long-run balanced growth paths are no longer parallel. In an interesting exchange, Lee, Pesaran and Smith (1998) criticize Islam (1995) for failing to allow for different time trends across countries. In response, Islam (1998) argues that Lee, Pesaran and Smith are assessing an economically uninteresting form of convergence when they allow for trend differences. This debate is an excellent example of the issues of interpretation that are raised in moving between specific economic hypotheses and more general statistical models. One drawback of many current panel studies is that the construction of the time series observations can appear arbitrary. There is no inherent reason why 5 or 10 years represent natural spans over which to average observations. Similarly, there is arbitrariness with respect to which time periods are aggregated. A useful endeavor would be the development of tools to ensure that panel findings are robust with respect to the assumptions employed in creating the panel from the raw data. More fundamentally, the empirical growth literature has not fully addressed the question of the appropriate time horizons over which growth models should be assessed. 61 The impact of controlling for fixed effects for interpreting β-convergence is recognized in the conclusion
to Islam (1995).
636
S.N. Durlauf et al.
For example, it remains unclear when business cycle considerations (or instances of output collapses) may be safely ignored when modeling the growth process. While cross-section studies that examine growth over 30–40 year periods might be exempt from this consideration, it is less clear that panel studies employing 5-year averages are genuinely informative about medium-run growth dynamics. 6.3. Event study approaches Although we have focused on the limitations of panel data methods, it is clear that the prospects for informative work of this kind should improve over time. The addition of further time periods is valuable in itself, and the history of developing countries in the 1980s and 1990s offers various events that introduce richer time series variation into the data. These events include waves of democratization, macroeconomic stabilization, financial liberalization, and trade liberalization, and panel data methods can be used to investigate their unfolding consequences for growth. An alternative approach has become popular, and proceeds in a similar way to event studies in the empirical finance literature. In event studies, researchers look for systematic changes in asset returns after a discrete event, such as a profits warning. In fields outside finance, before-and-after studies like this have proved an informative way to gauge the effects of devaluations [see Pritchett (2000a) for references], of inflation stabilization [Easterly (1996)] and the consequences of the debt crisis for investment, as in Warner (1992). Pritchett (2000a) argues that there is a great deal of scope for studying the growth impact of major events and policy changes in a similar way. The obvious approach is to study the time paths of variables such as output growth, investment and TFP growth, examined before and after such events. In empirical growth research, Henry (2000, 2003) has applied this form of analysis to the effects of stock market liberalization on investment and growth, Giavazzi and Tabellini (2004) have considered economic and political liberalizations, while Wacziarg and Welch (2003) have studied the effects of trade liberalization. Depending on the context, one can also study the response of other variables in a way that is informative about the channels of influence. For example, in the case of trade liberalization, it is natural to study the response of the trade share, as in the work of Wacziarg and Welch. The rigor of this method should not be overplayed. As with any other approach to empirical growth, one has to be cautious about inferring a causal effect. This is clear from exploring the analogy with treatment effects, a focus of recent research in microeconometrics and labor economics.62 In the study of growth, the treatments – such as democratization – are clearly not exogenously assigned, but are events that have arisen 62 This connection with the treatment effect literature is sometimes explicitly made, as in Giavazzi and
Tabellini (2004) and Persson and Tabellini (2003). The connection helps to understand the limitations of the evidence, but the scope for resolving the associated identification problems may be limited in cross-country data sets.
Ch. 8: Growth Econometrics
637
endogenously. Moreover, the treatment effects will be heterogeneous and could depend, for example, on whether a policy change is seen as temporary or permanent [Pritchett (2000a)]. In these circumstances, the ability to quantify even an average treatment effect is strongly circumscribed. It may be possible to identify the direction of effects, and here the limited number of observations does have one advantage. With a small number of cases to examine, it is easy for the researcher to present a graphical analysis that allows readers to gauge the extent of heterogeneity in responses, and the overall pattern. At the very least, this offers a useful complement to regression-based methods. 6.4. Endogeneity and instrumental variables A final set of data-based issues concerns the identification of instrumental variables in cross-section and time series contexts. An obvious and frequent criticism of growth regressions is that they do little to establish directions of causation. At one level, there is the standard problem that two variables may be correlated but jointly determined by a third. It is very easy to construct growth examples. Variables such as growth and political stability could be seen as jointly determined equilibrium outcomes associated with, say, a particular set of institutions. In this light, a correlation between growth and political stability, even if robust in statistical terms, does not appear especially informative about the structural determinants of growth. There are many instances in growth research when explanatory variables are clearly endogenously determined (in the economic, not the statistical sense). The most familiar example would be a regression that relates growth to the ratio of investment to GDP. This may tell us that the investment share and growth are associated, but stops short of identifying a causal effect. Even if we are confident that a change in investment would affect growth, in a sense this just pushes the relevant question further back, to an understanding of what determines investment. When variables are endogenously determined in the economic sense, there is also a strong chance that they will be endogenous in the technical sense, namely correlated with the disturbances in the structural equation for growth. To give an example, consider what happens if political instability lowers growth, but slower economic growth feeds back into political instability. The estimated regression coefficient will tend to conflate these two effects and will be an inconsistent estimate of the causal effect of instability.63 Views on the importance of these considerations differ greatly. One position is that the whole growth research project effectively capsizes before it has even begun, but Mankiw (1995) and Wacziarg (2002) have suggested an alternative view. According to them, one should accept that reliable causal statements are almost impossible to make, 63 Although this ‘reverse causality’ interpretation of endogeneity is popular and important, it should be re-
membered that a correlation between an explanatory variable and the error term can arise for other reasons, including omitted variables and measurement error. As we discuss, it is important to bear this more general interpretation of the error term in mind when judging the plausibility of exclusion restrictions in instrumental variable procedures.
638
S.N. Durlauf et al.
but use the partial correlations of the growth literature to rule out some possible hypotheses about the world. Wacziarg uses the example of the negative partial correlation between corruption and growth found by Mauro (1995). Even if shown to be robust, this correlation does not establish that somehow reducing corruption will be followed by higher growth rates. But it does make it harder to believe some of the earlier suggestions, rarely based on evidence, that corruption could be actively beneficial. One approach is to model as many as possible of the variables that are endogenously determined. A leading example is Tavares and Wacziarg (2001), who estimate structural equations for various channels through which democracy could influence development. In their analysis, democracy affects growth via factors such as its effect on human capital accumulation, physical capital accumulation, inequality and government expenditures. They conclude the net effect of democracy on growth is slightly negative, despite the positive contributions that are made from the role of democracy in promoting greater human capital and reduced inequality. This approach has some important advantages in both economic and statistical terms. It can be informative about underlying mechanisms in a way that much empirical growth research is not. From a purely statistical perspective, if the structural equations are estimated jointly by methods such as three stage least squares or full information maximum likelihood, this is likely to bring efficiency gains. That said, systems estimation is not necessarily the best route: it has the important disadvantage that specification errors in one of the structural equations could contaminate the estimates obtained for the others. The most common response to the endogeneity of growth determinants has been the application of instrumental variable procedures to a single structural equation, with growth as the dependent variable. As mentioned in Section 4, two growth studies that employ instrumental variables estimators based on lagged explanatory variables are Barro and Lee (1994) and Caselli, Esquivel and Lefort (1996). Appendices C and D describe a wide range of other instrumental variables that have been proposed for the Solow variables and other growth determinants respectively, where the focus has been on the endogeneity of particular variables. The variety of instruments that have been proposed illustrates that it is relatively straightforward to find an instrument that is correlated with the endogenous explanatory variable(s). This apparent success may be illusory. In our view, the belief that it is easy to identify valid instrumental variables in the growth context is deeply mistaken. We regard many applications of instrumental variable procedures in the empirical growth literature to be undermined by the failure to address properly the question of whether these instruments are valid, i.e., whether they may be plausibly argued to be uncorrelated with the error term in a growth regression. When the instrument is invalid, instrumental variables estimates will of course be inconsistent. Not enough is currently known about the consequences of “small” departures from validity, but it is certainly possible to envisage circumstances under which ordinary least squares would be preferable to instrumental variables on, say, a mean square error criterion. A common misunderstanding, perhaps based on confusing the economic and statistical versions of “exogeneity”, is that predetermined variables, such as geographical
Ch. 8: Growth Econometrics
639
characteristics, are inevitably strong candidates for instruments. There is, however, nothing in the predetermined nature of these variables to ensure either that they are not direct growth determinants or that they are uncorrelated with omitted growth determinants. Even if we take the extreme (from the perspective of being predetermined) example of geographic characteristics, there are many channels through which these could affect growth, and therefore many ways in which they could be correlated with the disturbances in a growth model. Brock and Durlauf (2001a) use this type of reasoning to make a very general critique of the use of instrumental variables in growth economics, basing it on the notion of theory open-endedness that we have described earlier. Since growth theories are mutually compatible, the validity of an instrument requires a positive argument that it cannot be a direct growth determinant or correlated with an omitted growth determinant. For many of the instrumental variables that have been proposed, this is clearly not the case. Discussions of the validity of instruments inevitably suffer from some degree of imprecision because of the need to make qualitative and subjective judgments. When one researcher claims that it is implausible that a given instrument is valid, unless this claim is made on the basis of a joint model of the instruments and the variable of original interest, another researcher can always simply reject the assertion as unpersuasive. To be clear, this element of subjectivity does not mean that arguments about validity are pointless.64 Rather, one must recognize that not all statistical questions can be adjudicated on the basis of mathematical analysis. To see how different instruments might be assigned different levels of plausibility, we consider two examples. Brock and Durlauf (2001a) single out Frankel and Romer’s (1999) geographic instruments as an example where instrument validity appears suspect as such variables are likely correlated with features of a country’s economic, political, legal, and social institutions.65 In our view, the large body of theoretical and empirical evidence on the role of institutions on growth, as well as even a cursory reading of history, renders the orthogonality assumptions required to use the instruments questionable.66 For example, it is a standard historical claim that the fact that Great Britain is an island had important implications for its political development. While Frankel 64 Put differently, one does not require a precise definition of what makes an instrument valid in order to
argue whether a given instrument is valid or not. To take an example due to Taylor (1998), the absence of a precise definition of money does not weaken my belief that the currency in my wallet is a form of money, whereas the computer on which this paper is written is not. To claim such arguments cannot be made is known as the Socratic fallacy. 65 While questions about the validity of instrumental variables arise in virtually all contexts, the force of these concerns differs across contexts. For example, in rational expectations models, lagged variables are natural instruments with respect to variables that, from the perspective of the theoretical model, are martingale differences, as occurs for excess holding returns. Objections to particular instruments in these contexts typically rely on alternative specifications of preferences or some other modification of the economic logic of the original model. This is quite different from the open-endedness of growth theories. 66 The body of work on institutions and growth excellently summarized in Acemoglu, Johnson and Robinson (2004) is supportive of this claim.
640
S.N. Durlauf et al.
(2003) suggests that this worry is contrived, the argument against instrument validity flows quite naturally from modern growth theory and the many possible ways in which geographic characteristics such as remoteness could influence development. As an example where instrument validity may be more plausible, consider Cook (2002a). He employs measures of damage caused by World War II as instruments for various growth regressors such as savings rates. The validity of Cook’s instruments again relies on the orthogonality of World War II damage with omitted postwar growth determinants. It may be that levels of wartime damage had consequences for post-War growth performance in other respects (such as institutional change) but this argument is perhaps less straightforward than in the case of geographic characteristics. To be clear, this discussion is nowhere near sufficient to conclude that Frankel and Romer’s instruments are invalid whereas Cook’s are valid. Rather our point is that conclusions concerning the relative plausibility of one set of instruments versus another need to rest on explicit arguments. It is not enough to appeal to a variable being predetermined, because this does not ensure that it is uncorrelated with the disturbances in the structural equation being estimated. A key implication of our discussion is that historical information has a vital role to play in facilitating formal growth analyses and evaluating exclusion restrictions. This discussion of instrumental variables indicates another important, albeit neglected, issue in empirical growth analysis: the relationship between model specification and instrumental variable selection. One cannot discuss the validity of particular instruments independently from the choice of the specific growth determinants under study. An important outstanding research question is whether model uncertainty and instrumental variable selection can be integrated simultaneously into some of the methods we have described, including model averaging and automated model selection. The recent work of Hendry and Krolzig (2005) on automated methods includes an ambitious approach to systematic model selection for simultaneous equation models in which identifying restrictions are determined by the data.
7. Econometric issues II: Data and error properties In this section we consider a range of questions that arise in growth econometrics from the properties of data and errors. Starting with data issues, Section 7.1 examines how one may handle outliers in growth data. Section 7.2 examines the problem of measurement error. This is an important issue since there are good reasons to believe that the quality of the data is sometimes poor for less developed economies. In Section 7.3 we consider the case where data are not even measured, i.e. are missing. Turning to issues of the properties of model errors, Section 7.4 examines the analysis of heteroskedasticity in growth contexts. Finally, Section 7.5 addresses the problem of cross-section correlation in model errors.
Ch. 8: Growth Econometrics
641
7.1. Outliers Empirical growth researchers often work with small data sets and estimate relatively simple models. In these circumstances, OLS regressions are almost meaningless unless they have been accompanied by systematic investigation of the data, including the sensitivity of the results to outlying observations. There are various reasons why some observations may be unrepresentative. It is possible for variables to be measured with error for that particular region or country. Alternatively, the model specified by the researcher may omit a relevant consideration, and so a group of country observations will act as outliers. By construction, least squares estimates can be highly sensitive to the presence of small groups of observations. The practical implication is that OLS can give a misleading account of the patterns in the majority of the data. The dangers of using OLS were forcibly expressed by Swartz and Welsch (1986, p. 171): “In a world of fat-tailed or asymmetric error distributions, data errors, and imperfectly specified models, it is just those data in which we have the least faith that often exert the most influence on the OLS estimates”. Some researchers respond to this concern using leverage measures or single-case diagnostics such as Cook’s distance statistic. There are well-known problems with these approaches, because where more than one outlier is present, its effect can easily be hidden by another (known in the statistics literature as “masking”). By far the best response is to use a more robust estimator, such as least trimmed squares, at least as a preliminary way of investigating the data.67 These issues are discussed in more detail in Temple (1998, 2000b). 7.2. Measurement error We now turn to a more general discussion of measurement error. It is clear that measurement errors are likely to be pervasive, especially in data that relate to developing countries. Concepts that appear straightforward in economic models can present huge measurement problems in practice, as in the example of the capital stock discussed by Pritchett (2000b). Yet relatively few empirical studies of growth consider the impact of measurement error in any detail. The best-known statistical result applies to a bivariate model where the independent variable is measured with error.68 The estimate of the slope coefficient will be biased towards zero, even in large samples, because measurement error induces covariance between the observable form of the regressor and the error term. This attenuation bias is
67 This estimator should not be confused with trimmed least squares, and other methods based on deleting
observations with high residuals in the OLS estimates. A residual-based approach is inadequate for obvious reasons. 68 This and the following discussion assume classical measurement error. Under more general assumptions, it is usually even harder to identify the consequences of measurement error for parameter estimates.
642
S.N. Durlauf et al.
well known, but sometimes misleads researchers into suggesting that measurement error will only mask effects, a claim that is not true in general. When there are multiple explanatory variables, but only one is measured with error, then typically all the parameter estimates will be biased. Some parameter estimates may be biased away from zero and, although the direction of the bias can be estimated consistently, this is rarely done. When several variables are measured with error, the assumption that measurement error only hides effects is even less defensible. Where measurement error is present, the coefficients are typically not identified unless other information is used. The most popular solution is to use instrumental variables, if an instrument can be found which is likely to be independent of the measurement error. A more complex solution is to exploit higher-order sample moments to construct more sophisticated estimators, as in Dagenais and Dagenais (1997). These procedures may be unreliable in small samples since the use of higher-order moments will make them especially sensitive to outliers. Sometimes partial identification is possible, in the sense that bounds on the extent of measurement error can be used to derive consistent estimates of bounds on the slope parameters. Although it can be difficult for researchers to agree on sensible bounds on the measurement error variances, there are easier ways of formulating the necessary restrictions, as discussed by Klepper and Leamer (1984). Their reverse regression approach was implemented by Persson and Tabellini (1994) and Temple (1998), but has rarely been used by other researchers. Another strategy is to investigate sensitivity to varying degrees of measurement error, based on method-of-moments corrections. Again, this is easy to implement in linear models, and should be applied more routinely than it is at present. Temple (1998) provides a discussion of both approaches in the context of estimating technology parameters and the rate of conditional convergence within the Mankiw, Romer and Weil (1992) model. 7.3. Missing data Some countries never appear in growth data sets, partly by design: it is common to leave out countries with very small populations, oil producers, and transition economies. These are countries that seem especially unlikely to lie on a regression surface common to the majority of the OECD countries or the developing world. Countries with small populations should not be allowed to carry a great deal of weight in attempting to draw generalizations about growth for larger countries. Other countries are left out for different reasons. When a nation experiences political chaos, or lacks economic resources, the collection of national accounts statistics will be a low priority. This means that countries like Afghanistan, Ethiopia and Somalia rarely appear in comparative growth studies. In other cases, countries appear in some studies but not in others, depending on the availability of particular variables of interest. Missing data are of course a potentially serious problem. If one started from a representative data set and then deleted countries at random, this would typically increase the standard errors but not lead to biased estimates. More serious difficulties arise if
Ch. 8: Growth Econometrics
643
countries are missing in a systematic way, because then parameter estimates are likely to be biased. This problem is given relatively little attention in mainstream econometrics textbooks, despite a large body of research in the statistics literature. A variety of solutions are possible, with the simplest being one form or another of imputation, with an appropriate adjustment to the standard errors. Hall and Jones (1999) and Hoover and Perez (2004) are among the few empirical growth studies to carry out imputation in a careful and systematic way. This approach may be especially useful when countries are missing from a data set because a few variables are not observed for their particular cases. It is then easy to justify using other available information to predict the missing data, and thereby exploit the additional information in the variables that are observed. Alternative approaches to missing data are also available, based on likelihood or Bayesian methods, which can be extended to handle missing observations. 7.4. Heteroskedasticity It is common in cross-section regressions for the underlying disturbances to have a non-constant variance. As is well known, the coefficient estimates remain unbiased, but OLS is inefficient and the estimates of the standard errors are biased. Most empirical growth research simply uses the heteroskedasticity-consistent standard errors developed by Eicker (1967) and White (1980). These estimates of the standard errors are consistent but not unbiased, which suggests that alternative solutions to the problem may be desirable. For data sets of the size found in cross-country empirical work, the alternative estimators developed by MacKinnon and White (1985) are likely to have better finite sample properties, as discussed in Davidson and MacKinnon (1993) and supported by simulations in Long and Ervin (2000). There are at least two other concerns with the routine application of White’s heteroskedasticity correction as the only response to heteroskedasticity. The first is that by exploiting any structure in the variance of the disturbances, using weighted least squares, it may be possible to obtain efficiency gains. The second and more fundamental objection is that heteroskedasticity can often arise from serious model misspecification, such as omitted variables or neglected parameter heterogeneity. Evidence of heteroskedasticity should then prompt revisions of the model for the conditional mean, rather than mechanical adjustments to the standard errors. See Zietz (2001) for further discussion and references. 7.5. Cross-section error correlation An unresolved issue in growth econometrics is the treatment of cross-section correlation in model errors. Such correlation may have important consequences for inference; as noted by DeLong and Summers (1991) in the growth context, failure to account for cross-sectional dependence can lead to incorrect calculation of standard errors and hence, incorrect inferences. One would certainly expect cross-sectional dependence to
644
S.N. Durlauf et al.
be present when studying growth. For example, countries that are geographically close together, or trading partners, may experience common shocks. Whether this effect is sizeable remains an open question, but one that might be addressed using ideas developed in Baltagi, Song and Koh (2003) and Driscoll and Kraay (1998), among others. In the context of growth regressions, work on cross-section dependence may be divided into two lines. One direction concerns the identification of the presence of cross-section dependence. Pesaran (2004b) develops tests for cross-section dependence that do not rely on any prior ordering; this framework in essence sums the cross-section sample error correlations in a panel and evaluates whether they are consistent with the null hypothesis that the population correlations are zero. Specifically, he proposes (recalling that N denotes the cross-section dimension and T the time dimension) a cross-section dependence statistic CD
N −1 N 2T CD = (66) ρˆi,j , N (N − 1) i=1 j =i+1
where ρˆi,j is the sample correlation between εi,t and εj,t ; Pesaran gives conditions under which this statistic converges to a Normal (0, 1) random variable (as N and T become infinite) under the null hypothesis of no cross-section correlation. This test statistic is based on earlier work by Breusch and Pagan (1980) and appears to possess good finite sample properties in comparison to this earlier work. Using a country-level panel, Pesaran (2004b) finds strong rejections of the null of no cross-section correlation both for the world as a whole as well as within several geographic groupings. The second and primary direction for the analysis of cross-section correlation has been concerned not so much with testing for its presence, but rather accounting for its presence in growth exercises. One approach relies on formulating a statistical model of the dependence. Phillips and Sul (2003) model the residuals in a growth panel as εi,t = δi θt + ui,t ,
(67)
where θt and ui,t are independent random variables; ui,t is assumed to be i.i.d. across countries and across time. Phillips and Sul (2002) describe the properties of panel estimators under this assumption. Another possibility in analyzing cross-sectional dependence is to treat the problem as one of spatial correlation in errors. The problem of spatial correlation has been much studied in the regional science literature, and statisticians in this field have developed spatial analogues of many time series concepts, see Anselin (2001) for an overview. Spatial methods have, in our view, an important role to play in growth econometrics. However, when these methods are adapted from the spatial statistics literature, they raise the problem of identifying the appropriate notion of space. One can imagine many reasons for cross-section correlation. If one is interested in technological spillovers, it may well be the case that in the space of technological proximity, the United Kingdom is closer to the United States than is Mexico. Put differently, unlike the time series
Ch. 8: Growth Econometrics
645
and spatial cases, there is no natural cross-section ordering to elements in the standard growth data sets. Following language due to Akerlof (1997) countries are perhaps best thought of as occupying some general socio-economic–political space defined by a range of factors; if one could identify their locations, then spatial methods could be implemented. An interesting approach to addressing the relevant spatial location of countries is pursued by Conley and Ligon (2002). In their analysis, they attempt to construct estimates of the spatial covariation of the residuals εi in a cross-section. In order to do this, they construct different measures of socioeconomic distance between countries. They separately consider geographic distance (measured between capital cities), as well as measures of the costs of transportation between these cities. Once a distance metric is constructed, these are used to construct a residual covariance matrix. Estimation methods for this procedure are developed in Conley (1999). Conley and Ligon (2002) find that allowing for cross-section dependence in this way is relatively unimportant in terms of appropriate calculation of standard errors for growth model parameters. Their methods could be extended to allow for comparisons of different variables as the source for cross-section correlation as is done in Conley and Topa (2002) in the context of residential neighborhoods. A valuable generalization of this work would be the modeling of cross-section correlations as a function of multiple variables. Such an analysis would make further progress on the measurement of distances in socioeconomic space, which, as we have suggested, presumably are determined by multiple channels. A generally unexplored possibility for studying cross-section dependence in growth (and other contexts) is to model these correlations structurally as the outcome of spillover effects.69 The theoretical literature on social interactions studies crosssectional dependence in precisely this way [see Brock and Durlauf (2001b) for a survey of this literature]. While such models have the potential for providing firm microfoundations for cross-section dependence, the presence of such spillovers has consequences for identification that are not easily resolved [Brock and Durlauf (2001b), Manski (1993)] and which have yet to be explored in growth contexts; Binder and Pesaran (2001) and Brock and Durlauf (2001b) analyze identification and estimation problems for intertemporal environments that are particularly germane to growth contexts.
8. Conclusions: The future of growth econometrics In this section, we offer some closing thoughts on the most promising directions for empirical growth research. We are not the first authors to set out manifestoes for the field, and we explicitly draw on previous contributions, many of which deserve wider currency. It is also interesting to compare the current state of the field against the verdicts offered in the early survey by Levine and Renelt (1991). One dominant theme will be
69 An exception is Easterly and Levine (1997b).
646
S.N. Durlauf et al.
that the empirical study of growth requires an eclectic approach, and that the field has been harmed by a tendency for research areas to evolve independently, without enough interaction.70 This is not simply a question of using a variety of techniques: it also means that there needs to be a closer connection between theory and evidence, a willingness to draw on ideas from areas such as trade theory, and more attention to particular features of the countries under study. We start with Pritchett (2000a), who lists three questions for growth researchers to address: • What are the conditions that initiate an acceleration of growth or the conditions that set off sustained decline? • What happens to growth when policies – trade, macroeconomic, investment – or politics change dramatically in episodes of reform? • Why have some countries absorbed and overcome shocks with little impact on growth, while others seem to have been overwhelmed by adverse shocks? This agenda seems to us very appropriate, not least because it focuses attention on substantive economic issues rather than the finer points of estimating aggregate production functions. The importance of the first of Pritchett’s questions is evident from the many instances where countries have moved from stagnation to growth and vice versa. A paper by Hausmann, Pritchett and Rodrik (2004) explicitly models transitions to fast growth (“accelerations”) and makes clear the scope for informative work of this kind. The second question we have discussed above, and research in this vein is becoming prominent, as in Henry (2000, 2003), Giavazzi and Tabellini (2004), and Wacziarg and Welch (2003). Here, one of the major challenges will be to relax the (sometimes only implicit) assumption that policies are randomly assigned. Finally, an important paper by Rodrik (1999) has addressed the third question, namely what determines varying responses to major shocks. In all three cases, it is clear that econometric work should be informed by detailed studies of individual countries, such as those collected in Rodrik (2003). Too much empirical growth research proceeds without enough attention to the historical and institutional context. For example, a newcomer to this literature might be surprised at the paucity of work that integrates growth regression findings with, say, the known consequences of the 1980s debt crisis. Another reason for advocating case studies is that much of the empirical growth literature essentially points only to reduced-form partial correlations. These can be useful, but it is clear that we often need to move beyond this. A partial correlation is more persuasive if it can be supported by theoretical arguments. The two combined are more persuasive if there is evidence of the intermediating effects or mechanisms that are emphasized in the relevant theory. There is plenty of scope for informative work that tries 70 To give a specific example, the macroeconomic literature on international technology differences only
rarely acknowledges relevant work by trade economists, including estimates of the Heckscher–Ohlin–Vanek model that suggest an important role for technology differences. See Klenow and Rodriguez-Clare (1997b) for more discussion.
Ch. 8: Growth Econometrics
647
to isolate mechanisms by which variables such as financial depth, inequality, and political institutions shape the growth process. Wacziarg (2002), in particular, highlights the need for a structural growth econometrics, one that aims to recover channels of causation, and hence supports (or undermines) the economic significance of the partial correlations identified in the literature. A more extreme view is that growth econometrics should be supplanted by the calibration of theoretical models. Klenow and Rodriguez-Clare (1997b) emphasize the potential of such an approach and note that Mankiw, Romer and Weil’s (1992) influential analysis can be seen partly as a comparison of estimated parameter values with those associated with specific theoretical models. Relatively little of the empirical work that has followed has achieved a similarly close connection between theory and evidence, and this has been a recurring criticism of the literature [for example, Levine and Renelt (1991) and Durlauf (2001)]. It may be premature to say that econometric approaches should be entirely replaced by calibration exercises, but the two methods could surely inform each other more often than at present. Calibrated models can help to interpret parameter estimates, not least in comparing the magnitude of the estimates with the implications of plausible models. Klenow and Rodriguez-Clare (1997b) discuss examples of this in more detail. At the same time, the partial correlations identified in growth econometrics can help to act as a discipline on model-building and can indicate where model-based quantitative investigations are most likely to be fruitful. This role for growth econometrics is likely to be especially useful in areas where the microeconomic evidence used to calibrate structural models is relatively weak, or the standard behavioral assumptions may be flawed. The need for a tighter connection between theory and evidence is especially apparent in certain areas. The workhorse model for many empirical growth papers continues to be Solow–Swan, a closed economy model which leaves out aspects of interdependence that are surely important. Howitt (2000) has shown that growth regression evidence can be usefully reinterpreted in the light of a multi-country theoretical model with a role for technology diffusion. More generally, there is a need for researchers to develop empirical growth frameworks that acknowledge openness to flows of goods, capital and knowledge. These issues are partly addressed by the theoretical analysis of Barro, Mankiw and Sala-i-Martin (1995) and empirical work that builds on such ideas deserves greater prominence. Here especially, research that draws on the quantitative implications of specific models, as in the work of Eaton and Kortum (1999, 2001) on technology diffusion and the role of imported capital goods, appears to be an important advance. The neglect of open economy aspects of the countries under study is mirrored elsewhere. Much of the empirical literature uses a theoretical framework that was originally developed to explain the growth experiences of the USA and other developed nations. Yet this framework is routinely applied to study developing countries, and there appears plenty of scope for models that incorporate more of the distinctive features of poorer countries. These could include the potentially important roles of agricultural employment, dualism, and structural change, and in some cases, extensive state involvement
648
S.N. Durlauf et al.
in production. This is an area in which empirical growth researchers have really only scratched the surface. Some of these issues are connected to an important current research agenda, namely the need to distinguish between different types of growth and their distributional consequences. For example, the general equilibrium effects of productivity improvements in agriculture may be very different to those in services and industry. Identifying the nature of “pro-poor” growth will require more detailed attention to particular features of developing countries. Given that the main source of income for the poor is usually labour income, growth researchers will need to integrate their models with theory and evidence from labour economics, in order to study how growth and labour markets interact. Agénor (2004) considers some of the relevant issues, and again this appears to be a vital direction for future research. Ideally, research along these various lines will utilize not only statistics, but also the power of case studies in generating hypotheses, and in deepening our understanding of the economic, social and political forces at work in determining growth outcomes. Case studies may be especially valuable in two areas. The first of these is the study of technology transfer. As emphasized in the survey by Klenow and Rodriguez-Clare (1997b), we do not know enough about why some countries are more successful than others in climbing the “ladder” of product quality and technological complexity. What are the relative contributions of human capital, foreign direct investment and trade? In recent years some of these issues have been intensively studied at the microeconomic level, especially the role of foreign direct investment and trade, but there remains work to be done in mapping firm and sector-level evidence into a set of aggregate implications. A second area in which case studies are likely to prove valuable is the study of political economy, in its modern sense. It is a truism that economists, particularly those considering development, have become more aware of the need to account for the twoway interaction between economics and politics. A case can be made that the theoretical literature has outpaced the empirical literature in this regard. Studies of individual countries, drawing on both economic theory and political science, would help to close this gap. Thus far, we have highlighted a number of limitations of existing work, and directions in which further research seems especially valuable. Some of the issues we have considered were highlighted much earlier by Levine and Renelt (1991), and that might lead to pessimism over the long-term prospects of this literature.71 This also shows that our prescriptions for future research could seem rather pious, since the improvements we recommend are easier said than done. We end our review by considering some areas in which genuine progress has been made, and where further progress appears likely.
71 Only now are researchers beginning to engage with some of the issues they raised, such as the varying
conditions under which it is appropriate to use international rather than national prices in making productivity comparisons and constructing capital stocks.
Ch. 8: Growth Econometrics
649
One reason for optimism is the potential that recently developed model averaging methods have for shedding new light on growth questions. These methods help to address the model selection and robustness issues that have long been identified as a major weakness of cross-country growth research. By framing the problem explicitly in terms of model uncertainty, in the way envisaged by Leamer (1978), it is possible to consider many candidate explanatory variables simultaneously, and identify which effects appear to be systematic features of the data, as reflected in posterior probabilities of inclusion. The Bayesian approach to model averaging also provides an index of model adequacy, the posterior model probability, that is easy to interpret, and that allows researchers to gauge the extent of overall model uncertainty. Above all, researchers can communicate the degree of support for a particular hypothesis with more faith that the results do not depend on an arbitrary choice of regression specification. Although the application of Bayesian model averaging inevitably has limitations of its own, it appears more rigorous than many of the alternatives, and we expect a number of familiar growth questions to be revisited using these methods. Another reason for optimism is that the quality of available data is likely to improve over time. The development of new and better data has clearly been one of the main achievements of the empirical growth literature since the early 1990s, and one that was not foreseen by critics of the field. Researchers have developed increasingly sophisticated proxies for drivers of growth that appeared resistant to statistical analysis. One approach, pioneered in the growth literature by Knack and Keefer (1995) and Mauro (1995), has been country-specific ratings compiled by international agencies. Such data increasingly form the basis for measures of corruption, government efficiency, and protection of property rights. More recent work such as that of Kaufmann, Kraay and Zoido-Lobaton (1999a, 1999b) and Kaufmann, Kraay and Mastruzzi (2003) has established unusually comprehensive measures of various aspects of institutions. The construction of proxies is likely to make increasing use of latent variable methods. These aim to reduce a set of observed variables to a smaller number of indicators that are seen as driving the majority of the variation in the original data, and that could represent some underlying variable of interest. For example, the extent of democracy is not directly observed, but is often obtained by applying factor analysis or extracting principal components from various dimensions of political freedom. There are obvious dangers with this approach, but the results can be effective proxies for concepts that are otherwise hard to measure.72 They also help to overcome the dimensionality problem associated with cross-country empirical work. To be successfully employed, the rigorous use of a latent variable as a regressor will generally need to acknowledge the presence of measurement error.73 72 A relevant question, not often asked, is how high the correlation between the proxy and the true predictor
has to be for the estimated regression coefficient on the proxy to be of the “true” sign. Krasker and Pratt (1986, 1987) have developed methods that can be used to establish this under surprisingly general assumptions. 73 In principle this can be addressed by structural equation modeling, using software like EQS or LISREL to estimate a system of equations that includes explicit models for latent variables, an approach used elsewhere
650
S.N. Durlauf et al.
Using latent variables makes especially good sense under one view of the proper aims of growth research. It is possible to argue that empirical growth studies will never give good answers to precise hypotheses, but can be informative at a broader level. For example, a growth regression is unlikely to tell us whether the growth effect of inflation is more important than the effect of inflation uncertainty, because these two variables are usually highly correlated. It may even be difficult to distinguish the effects of inflation from the effects of sizeable budget deficits.74 Instead a growth regression might be used to address a less precise hypothesis, such as the growth dividend of macroeconomic stability, broadly conceived. In this context, it is natural to use latent variable approaches to measure the broader concept. Another valuable development is likely to be the creation of rich panel data sets at the level of regions within countries. Regional data offer greater scope for controlling for some variables that are hard to measure at the country level, such as cultural factors. By comparing experiences across regions, there may also be scope for identifying events that correspond more closely to natural experiments than those found in cross-country data. Work such as that by Besley and Burgess (2000, 2002, 2004) using panel data on Indian states shows the potential of such an approach. In working with such data more closely, one of the main challenges will be to develop empirical frameworks that incorporate movements of capital and labour between regions: clearly, regions within countries should only rarely be treated as closed economies. Shioji (2001b) is an example of how analysis using regional data can take this into account. Even with better data, at finer levels of disaggregation, the problem of omitted variables can only be alleviated, not resolved. It is possible to argue that the problem applies equally to historical research and case studies, but at least in these instances, the researcher may have some grasp of important forces that are difficult to quantify. Since growth researchers naturally gravitate towards determinants of growth that can be analyzed statistically, there is an ever-present danger that the empirical literature, even taken as a whole, yields a rather partial and unbalanced picture of the forces that truly matter. Even a growth model with high explanatory power, in a statistical sense, has to be seen as a rather provisional set of ideas about the forces that drive growth and development. This brings us to our final points. We once again emphasize that empirical progress on the major growth questions requires attention to the evidence found in qualitative sources such as historical narratives and studies by country experts. One example we have given in the text concerns the validity of instrumental variables: understanding the historical experiences of various countries seems critical for determining whether
in the social sciences. Most economists are not familiar with this approach, and this makes the assumptions and results hard to communicate. It is therefore not clear that a full latent variable model should be preferred to a simpler solution, such as one of those we discuss in the measurement error section above. 74 As Sala-i-Martin (1991) has argued, various specific indicators of macroeconomic instability should perhaps be seen as symptoms of some deeper, underlying characteristic of a country.
Ch. 8: Growth Econometrics
651
exclusion restrictions are plausible. In this regard work such as that of Acemoglu, Johnson and Robinson (2001, 2002) is exemplary. More generally, nothing in the empirical growth literature suggests that issues of long-term development can be disassociated from the historical and cultural factors that fascinated commentators such as Max Weber. Where researchers have revisited these issues, as in Barro and McCleary (2003), the originality resides less in the conception of growth determinants and more in the scope for new statistical evidence. Of course, the use of historical analysis also leads back to the value of case studies, a point that has recurred throughout this discussion. In conclusion, growth econometrics is an area of research that is still in its infancy. To its credit, the field has evolved in response to the substantive economic questions that arise in growth contexts. The nature of the field has also led econometricians to introduce a number of statistical methods into economics, including classification and regression tree algorithms, robust estimation, threshold models and Bayesian model averaging, that appear to have wide utility. As with any new literature, especially one tackling questions as complex as these, it is possible to identify significant limitations of the existing evidence and the tools that are currently applied. But the progress that has been made in growth econometrics in the brief time since its emergence gives reason for continued optimism.
Acknowledgements Durlauf thanks the University of Wisconsin and John D. and Catherine T. MacArthur Foundation for financial support. Johnson thanks the Department of Economics, University of Wisconsin for its hospitality in Fall 2003, during which part of this chapter was written. Temple thanks the Leverhulme Trust for financial support under the Philip Leverhulme Prize Fellowship scheme. Ritesh Banerjee, Ethan Cohen-Cole, Giacomo Rondina and Lisa Wong have provided excellent research assistance. Finally, we thank Gordon Anderson, William Brock and Stephen Bond for useful discussions.
Appendix A: Data Key to the 102 countries AGO, Angola; ARG, Argentina; AUS, Australia; AUT, Austria; BDI, Burundi; BEL, Belgium; BEN, Benin; BFA, Burkina Faso; BGD, Bangladesh; BOL, Bolivia; BRA, Brazil; BWA, Botswana; CAF, Central African Republic; CAN, Canada; CHE, Switzerland; CHL, Chile; CHN, China; CIV, Cote d’Ivoire; CMR, Cameroon; COG, Rep. of Congo; COL, Colombia; CRI, Costa Rica; CYP, Cyprus; DNK, Denmark; DOM, Dominican Republic; ECU, Ecuador; EGY, Egypt; ESP, Spain; ETH, Ethiopia; FIN, Finland; FJI, Fiji; FRA, France; GAB, Gabon; GBR, United Kingdom; GHA, Ghana; GIN, Guinea; GMB, The Gambia; GNB, Guinea-Bissau; GRC, Greece; GTM, Guatemala;
652
S.N. Durlauf et al.
GUY, Guyana; HKG, Hong Kong; HND, Honduras; IDN, Indonesia; IND, India; IRL, Ireland; IRN, Iran; ISR, Israel; ITA, Italy; JAM, Jamaica; JOR, Jordan; JPN, Japan; KEN, Kenya; KOR, Rep. of Korea; LKA, Sri Lanka; LSO, Lesotho; MAR, Morocco; MDG, Madagascar; MEX, Mexico; MLI, Mali; MOZ, Mozambique; MRT, Mauritania; MUS, Mauritius; MWI, Malawi; MYS, Malaysia; NAM, Namibia; NER, Niger; NGA, Nigeria; NIC, Nicaragua; NLD, Netherlands; NOR, Norway; NPL, Nepal; NZL, New Zealand; PAK, Pakistan; PAN, Panama; PER, Peru; PHL, Philippines; PNG, Papua New Guinea; PRT, Portugal; PRY, Paraguay; ROM, Romania; RWA, Rwanda; SEN, Senegal; SGP, Singapore; SLV, El Salvador; SWE, Sweden; SYR, Syria; TCD, Chad; TGO, Togo; THA, Thailand; TTO, Trinidad & Tobago; TUR, Turkey; TWN, Taiwan; TZA, Tanzania; UGA, Uganda; URY, Uruguay; USA, USA; VEN, Venezuela; ZAF, South Africa; ZAR, Dem. Rep. Congo; ZMB, Zambia; ZWE, Zimbabwe. Extrapolation Where data on GDP per worker for the year 2000 are missing from PWT 6.1, but are available for 1996 or after, we extrapolate using the growth rate between 1990 and the latest available year. This procedure helps to alleviate the biases that can occur when countries are missing from the sample for systematic reasons, such as political or economic collapse. The countries involved are Angola (extrapolated from 1990–1996), Botswana (1999), Central African Republic (1998), Democratic Republic of Congo (1997), Cyprus (1996), Fiji (1999), Guyana (1999), Mauritania (1999), Namibia (1999), Papua New Guinea (1999), Singapore (1996), and Taiwan (1998).
Appendix B: Variables in cross-country growth regressions
R.H.S. variables
Studies
Capitalism Capital account liberalization Corruption
Hall and Jones (1999) (+, ∗ ) Eichengreen and Leblang (2003) (+, ∗ ) Mauro (1995) (−, ∗ ) Welsch (2003) (−, ∗ )
Democracy Minimum levels . . . Higher levels Overall ‘Voice’ Demographic characteristics Share of population 15 or below Share of population 65 or over Growth of 15–65 population share
Barro (1996, 1997) (+, ∗ ) Barro (1996, 1997) (−, ∗ ) Alesina et al. (1996) (?, _ ) Minier (1998) (+, ∗ ) Dollar and Kraay (2003) (−, ∗ ) Barro and Lee (1994) (−, ∗ ) Barro and Lee (1994) (?, _ ) Bloom and Sachs (1998) (+, ∗ )
Ch. 8: Growth Econometrics
R.H.S. variables Education College level Female (level)
Female (growth) Male (level)
Male (growth) Overall (level)
Primary level Secondary level Initial income ∗ male schooling Proportion of engineering students Proportion of law students Ethnicity and language Ethno-linguistic fractionalization
Language diversity Fertility
Finance Stock markets
Banks Dollarization Depth
Competition ∗ development Repression Sophistication
653
Studies
Barro and Lee (1994) (−, _ ) Barro and Lee (1994) (−, ∗ ) Barro (1996, 1997) (−, ∗ ) Caselli, Esquivel and Lefort (1996) (+, ∗ ) Forbes (2000) (−, ∗ ) Barro and Lee (1994) (−, ∗ ) Barro and Lee (1994) (+, ∗ ) Barro (1996) (+, ∗ ) Caselli, Esquivel and Lefort (1996) (−, ∗ ) Forbes (2000) (+, ∗ ) Barro and Lee (1994) (+, ∗ ) Azariadis and Drazen (1990) (+, ∗ ) Barro (1991) (+, ∗ ) Knowles and Owen (1995) (+, _ ) Easterly and Levine (1997a) (+, ∗ ) Krueger and Lindahl (2001) (+, ∗ ) Bils and Klenow (2000) (+, ∗ ) Sachs and Warner (1995) (+, _ ) Barro (1997) (−, _ ) Sachs and Warner (1995) (+, _ ) Barro (1997) (−, ∗ ) Murphy, Shleifer and Vishny (1991) (+, ∗ ) Murphy, Shleifer and Vishny (1991) (−, ∗ ) Easterly and Levine (1997a) (−, ∗ ) Sala-i-Martin (1997a, 1997b) (?, _ ) Alesina et al. (2003) (−, ∗ ) Masters and McMillan (2001) (−, ∗ /_ ) Barro (1991, 1996, 1997) (−, ∗ ) Barro and Lee (1994) (−, ∗ ) Levine and Zervos (1998) (+, ∗ ) Bekaert, Harvey and Lundblad (2001) (+, ∗ ) Beck and Levine (2004) (+, ∗ ) Beck and Levine (2004) (+, ∗ ) Edwards and Magendzo (2003) (+, _ ) Berthelemy and Varoudakis (1996) (+, ∗ ) Odedokun (1996) (+, ∗ ) Ram (1999) (+, _ ) Rousseau and Sylla (2001) (+, ∗ ) Deidda and Fattouh (2002) (+, _ ) Demetriades and Law (2004) (+, ∗ ) Claessens and Laeven (2003) (+, ∗ ) Roubini and Sala-i-Martin (1992) (−, ∗ ) Easterly (1993) (−, ∗ ) King and Levine (1993) (+, ∗ ) Levine and Zervos (1993) (+, robust)
654
S.N. Durlauf et al.
R.H.S. variables
Studies Easterly and Levine (1997a) (+, ∗ ) Sala-i-Martin (1997a, 1997b) (?, _ )
Credit Growth rate Volatility Foreign direct investment Fraction of mining in GDP Geography Absolute latitude
Disease ecology
Frost days Land locked Coastline (length)
Arable land Rainfall Variance of rainfall Maximum temperature Government Consumption (growth) Consumption (level)
Deficits
Investment
Various expenditures Military expenditures
Levine and Renelt (1992) (+, not robust) De Gregorio and Guidotti (1995) (+, ∗ ) Levine and Renelt (1992) (+, not robust) Blonigen and Wang (2004) (+, _ ) Hall and Jones (1999) (+, ∗ ) Sala-i-Martin (1997a, 1997b) (+, ∗ ) Bloom and Sachs (1998) (+, ∗ ) Masters and McMillan (2001) (−, _ ) Easterly and Levine (2001) (+, ∗ ) Rodrik, Subramanian and Trebbi (2004) (+, ∗ ) McCarthy, Wolf and Wu (2000) (+, ∗ ) McArthur and Sachs (2001) (+, ∗ ) Easterly and Levine (2001) (−, ∗ ) Sachs (2003) (−, ∗ ) Masters and McMillan (2001) (+, ∗ ) Masters and Sachs (2001) (+, ∗ ) Easterly and Levine (2001) (−, ∗ ) Bloom and Sachs (1998) (+, ∗ ) Masters and Sachs (2001) (+, ∗ ) Bloom, Canning and Sevilla (2003) (+, ∗ ) Masters and Sachs (2001) (+, ∗ ) Masters and Sachs (2001) (+, ∗ ) Bloom, Canning and Sevilla (2003) (+, ∗ ) Bloom, Canning and Sevilla (2003) (−, ∗ ) Bloom, Canning and Sevilla (2003) (−, ∗ ) Kormendi and Meguire (1985) (+, _ ) Barro (1991) (−, ∗ ) Sachs and Warner (1995) (−, ∗ ) Barro (1996) (−, ∗ ) Caselli, Esquivel and Lefort (1996) (+, ∗ ) Barro (1997) (−, ∗ ) Acemoglu, Johnson and Robinson (2002) (−, _ ) Levine and Renelt (1992) (−, not robust) Fischer (1993) (−, ∗ ) Nelson and Singh (1994) (+, _ ) Easterly and Levine (1997a) (−, ∗ ) Bloom and Sachs (1998) (+, ∗ ) Barro (1991) (+, _ ) Sala-i-Martin (1997a, 1997b) (?, _ ) Kelly (1997) (+, ∗ ) Levine and Renelt (1992) (−, not robust) Aizenman and Glick (2003) (−, ∗ ) Guaresma and Reitschuler (2003) (−, ∗ )
Ch. 8: Growth Econometrics
655
R.H.S. variables
Studies
Military expenditures under threat Various taxes Growth rate of the G-7 countries in the previous period
Aizenman and Glick (2003) (+, ∗ ) Levine and Renelt (1992) (?, not robust)
Health Life expectancy
Change in malaria infection rate Adult survival rate Industrial structure % Small and medium enterprises Ease of entry and exit Inequality Democratic countries Non-democratic countries Overall
Inflation Growth Level
Variability
Infrastructure proxies
Initial income
Investment ratio
Alesina et al. (1996) (+, ∗ ) Easterly et al. (1993) (+, _ ) Alesina et al. (1996) (+, ∗ /_ ) Bloom, Canning and Sevilla (2004) (+, ∗ ) Barro and Lee (1994) (+, ∗ ) Bloom and Malaney (1998) (+, ∗ ) Bloom and Sachs (1998) (+, ∗ ) Bloom and Williamson (1998) (+, ∗ ) Hamoudi and Sachs (2000) (+, ∗ ) Gallup, Mellinger and Sachs (2000) (+, ∗ ) Gallup, Mellinger and Sachs (2000) Bhargava et al. (2001) Beck, Demirguc-Kunt and Levine (2003) (+, _ ) Beck, Demirguc-Kunt and Levine (2003) (+, ∗ ) Persson and Tabellini (1994) (−, ∗ ) Persson and Tabellini (1994) (+, _ ) Alesina and Rodrik (1994) (−, ∗ ) Forbes (2000) (+, ∗ ) Knowles (2001) (−, ∗ ) Kormendi and Meguire (1985) (−, ∗ ) Levine and Renelt (1992) (−, not robust) Levine and Zervos (1993) (?, not robust) Barro (1997) (−, ∗ ) (in the range above 15%) Bruno and Easterly (1998) (−, ∗ ) Motley (1998) (−, ∗ ) Li and Zou (2002) (−, ∗ ) Levine and Renelt (1992) (−, not robust) Fischer (1993) (−, ∗ ) Barro (1997) (+, _ ) Sala-i-Martin (1997a, 1997b) (?, _ ) Hulten (1996) (+, ∗ ) Easterly and Levine (1997a) (+, ∗ ) Esfahani and Ramirez (2003) (+, ∗ ) Kormendi and Meguire (1985) (−, ∗ ) Barro (1991) (−, ∗ ) Sachs and Warner (1995) (−, ∗ ) Harrison (1996) (?, _ ) Barro (1997) (−, ∗ ) Easterly and Levine (1997a) Barro (1991) (+, ∗ ) Barro and Lee (1994) (+, ∗ )
656
S.N. Durlauf et al.
R.H.S. variables
Studies Sachs and Warner (1995) (+, ∗ ) Barro (1996) (+, _ ) Caselli, Esquivel and Lefort (1996) (+, ∗ ) Barro (1997) (+, _ )
Investment type Equipment or fixed capital
Non-equipment Labor Productivity growth Productivity quality Labor force part. rate Luck External debt dummy External transfers Improvement in terms of trade
Money growth Neighboring countries’ education proxies, initial incomes, investment ratios and population growth rates Political instability proxies
Political rights and civil liberties indices Civil liberties
Overall Political rights
Political institutions Constraints on executive Judicial independence Property rights ICRG index Expropriation risk
DeLong and Summers (1993) (+, ∗ ) Blomstrom, Lipsey and Zejan (1996) (−, _ ) Sala-i-Martin (1997a, 1997b) (+, ∗ ) DeLong and Summers (1991) (+, ∗ ) Lichtenberg (1992) (+, ∗ ) Hanushek and Kimko (2000) (+, ∗ ) Blomstrom, Lipsey and Zejan (1996) (+, ∗ ) Easterly et al. (1993) (−, _ ) Easterly et al. (1993) (mixed, _ ) Easterly et al. (1993) (+, ∗ ) Fischer (1993) (+, ∗ ) Barro (1996) (+, ∗ ) Caselli, Esquivel and Lefort (1996) (+, ∗ ) Barro (1997) (+, ∗ ) Blattman, Hwang and Williamson (2003) (+, ∗ ) Coulombe and Lee (1995) (+, ∗ ) Kormendi and Meguire (1985) (+, _ ) Ciccone (1996) (+, ∗ ) Barro (1991) (−, ∗ ) Barro and Lee (1994) (−, ∗ ) Sachs and Warner (1995) (−, _ ) Alesina et al. (1996) (−, ∗ ) Caselli, Esquivel and Lefort (1996) (−, ∗ ) Easterly and Levine (1997a) (−, ∗ ) Kormendi and Meguire (1985) (+, _ ) Levine and Renelt (1992) (?, not robust) Barro and Lee (1994) (−, ∗ ) Sachs and Warner (1995) (+, ∗ ) Barro (1991) (?, _ ) Barro and Lee (1994) (+, ∗ ) Sala-i-Martin (1997a, 1997b) (+, ∗ ) Acemoglu, Johnson and Robinson (2001) (+, ∗ ) Feld and Voigt (2003) (+, ∗ ) Knack (1999) (+, ∗ ) Acemoglu, Johnson and Robinson (2001) (+, ∗ ) McArthur and Sachs (2001) (+, ∗ )
Ch. 8: Growth Econometrics
R.H.S. variables Population Density Growth
Price distortions Consumption price Investment price Price levels Consumption price Investment price Real exchange rate Black market premium
Distortions
Variability Regional effects Absolute latitude East Asia dummy Former Spanish colonies dummy Latin America dummy
Sub-Saharan Africa dummy
Religion Buddhist Catholic
657
Studies
Sachs and Warner (1995) (+, _ ) Kormendi and Meguire (1985) (−, ∗ ) Levine and Renelt (1992) (−, not robust) Mankiw, Romer and Weil (1992) (−, ∗ ) Barro and Lee (1994) (+, _ ) Kelley and Schmidt (1995) (−, ∗ ) Bloom and Sachs (1998) (−, ∗ ) Easterly (1993) (+, _ ) Harrison (1996) (−, ∗ ) Barro (1991) (−, ∗ ) Easterly (1993) (−, ∗ ) Easterly (1993) (+, _ ) Easterly (1993) (−, ∗ ) Sachs and Warner (1995) (−, ∗ ) Levine and Renelt (1992) (−, not robust) Barro and Lee (1994) (−, ∗ ) Barro (1996) (−, ∗ ) Harrison (1996) (−, ∗ ) Easterly and Levine (1997a) (−, ∗ ) Sala-i-Martin (1997a, 1997b) (−, ∗ ) Dollar (1992) (−, ∗ ) Easterly (1993) (−, _ ) Harrison (1996) (−, _ ) Sala-i-Martin (1997a, 1997b) (−, ∗ ) Acemoglu, Johnson and Robinson (2002) (−, _ ) Dollar (1992) (−, ∗ ) Barro (1996) (+, ∗ ) Barro and Lee (1994) (+, _ ) Barro (1997) (+, _ ) Barro (1996) (−, ∗ ) Barro (1991) (−, ∗ ) Barro and Lee (1994) (−, ∗ ) Barro (1997) (−, _ ) Easterly and Levine (1997a) (−, ∗ ) Sala-i-Martin (1997a, 1997b) (−, ∗ ) Barro (1991) (−, ∗ ) Barro and Lee (1994) (−, ∗ ) Barro (1997) (−, _ ) Easterly and Levine (1997a) (−, ∗ ) Sala-i-Martin (1997a, 1997b) (−, ∗ ) Barro (1996) (+, ∗ ) Sala-i-Martin (1997a, 1997b) (−, ∗ )
658
S.N. Durlauf et al.
R.H.S. variables
Confucian Muslim
Protestant
Religious belief Attendance Rule of law indices
Scale effects Total area Total labor force Social capital and related Social “infrastructure” Citizen satisfaction with government Civic participation Groups – as defined by Putnam, Leonardi and Nanetti (1993) Groups – as defined by Olson (1982) Institutional performance Civic community (index of participation, newspaper readership, political behavior) Trust
Social development index Extent of mass communication Kinship Mobility Middle class Outlook Social capital (WVS) Social capital (WVS) Social achievement norm
Studies Masters and Sachs (2001) (+, ∗ ) Barro (1996) (+, ∗ ) Barro (1996) (+, ∗ ) Sala-i-Martin (1997a, 1997b) (+, ∗ ) Masters and Sachs (2001) (+, _ ) Barro (1996) (+, ∗ ) Sala-i-Martin (1997a, 1997b) (−, ∗ ) Masters and Sachs (2001) (+, ∗ ) Barro and McCleary (2003) (+, ∗ ) Barro and McCleary (2003) (−, ∗ ) Barro (1996) (+, ∗ ) Acemoglu, Johnson and Robinson (2001) (+, ∗ ) Easterly and Levine (2001) (−, ∗ ) Dollar and Kraay (2003) (+, _ ) Alcalá and Ciccone (2004) (+, _/∗ ) Rodrik, Subramanian and Trebbi (2004) (+, ∗ ) Barro and Lee (1994) Sala-i-Martin (1997a, 1997b) (?, _ ) Barro and Lee (1994) Sala-i-Martin (1997a, 1997b) (?, _ ) Hall and Jones (1999) (+, ∗ ) Helliwell and Putnam (2000) (+, ∗ ) (within Italy) Helliwell (1996) (−, ∗ ) (within Asia) Knack and Keefer (1997) (+, ∗ ) Keefer and Knack (1997) (−, _ ), Keefer and Knack (1997) (+, _ ), Helliwell and Putnam (2000) (+, ∗ ) (Italy) Helliwell and Putnam (2000) (+, ∗ ) (Italy) Granato, Inglehart and Leblang (1996) (+, ∗ ) Helliwell (1996) (−, ∗ ) (Asia) Knack and Keefer (1997) (+, ∗ ), La Porta et al. (1997) (+, ∗ ) Beugelsdijk and van Schalk (2001) (+, _ ) Zak and Knack (2001) (+, ∗ ) Temple and Johnson (1998) Temple and Johnson (1998) Temple and Johnson (1998) Temple and Johnson (1998) Temple and Johnson (1998) Temple and Johnson (1998) Rupasingha, Goetz and Freshwater (2000) (+, ∗ ) Whiteley (2000) (+, ∗ ) Granato, Inglehart and Leblang (1996) (+, ∗ )
Ch. 8: Growth Econometrics
R.H.S. variables
Capability Trade policy indices Import penetration Leamer’s intervention index Years open 1950–1990 Openness indices (growth) Openness indices (level)
Outward orientation Tariff Liberalization Trade statistics Fraction of export/import/total trade in GDP
Fraction of primary products in total exports Growth in export–GDP ratio
FDI inflows relative to GDP Machinery and equipment imports Volatility of shocks Growth innovations Monetary shock War Casualties per capita Dummy
Duration
659
Studies Swank (1996) (−, ∗ ) Temple and Johnson (1998) (+, ∗ ) Levine and Renelt (1992) (?, not robust) Levine and Renelt (1992) (−, not robust) Sachs and Warner (1996) (+, ∗ ) Sala-i-Martin (1997a, 1997b) (+, ∗ ) Harrison (1996) (+, ∗ ) Levine and Renelt (1992) (?, not robust) Sachs and Warner (1995) (+, ∗ ) Harrison (1996) (+, ∗ ) Wacziarg and Welch (2003) (+, ∗ ) Levine and Renelt (1992) (?, not robust) Sala-i-Martin (1997a, 1997b) (?, _ ) Barro and Lee (1994) (−, _ ) Sala-i-Martin (1997a, 1997b) (?, _ ) Ben-David (1993, 1996) (?, _ ) Levine and Renelt (1992) (+, not robust) Easterly and Levine (1997a) (?, _ ) Frankel and Romer (1999) (+, ∗ ) Dollar and Kraay (2003) (+, _ ) Alcalá and Ciccone (2004) (+, ∗ ) Rodrik, Subramanian and Trebbi (2004) (+, _ ) Sachs and Warner (1996) (−, ∗ ) Sala-i-Martin (1997a, 1997b) (−, ∗ ) Feder (1982) (+, ∗ ) Kormendi and Meguire (1985) (+, ∗ ) 20+ studies others Blomstrom, Lipsey and Zejan (1996) Romer (1993) (+, ∗ ) Kormendi and Meguire (1985) (−, ∗ ) Ramey and Ramey (1995) (−, ∗ ) Kormendi and Meguire (1985) (−, ∗ ) Easterly et al. (1993) (−, _ ) Barro and Lee (1994) (−, _ ) Easterly and Levine (1997a) (?, _ ) Sala-i-Martin (1997a, 1997b) (−, ∗ ) Barro and Lee (1994) (+, _ )
+/− = sign of coefficient in the corresponding growth regression. ? = sign not reported. * = claimed to be significant.
_ = claimed to be insignificant.
660
S.N. Durlauf et al.
Appendix C: Instrumental variables for Solow growth determinants
Variable
Instrument
Study
GDP growth
Rainfall variation
Miguel, Satyanath and Segenti (2003)
GDP – initial
Lagged values
Barro and Sala-i-Martin (2004)
GDP – initial (per capital stock)
Newsprint consumption, and number of radios
Romer (1990)
GDP – initial
Log population initial and trade measure
Bosworth and Collins (2003)
Human capital
Natural disasters
Toya, Skidmore and Robertson (2003)
Investment – equipment
Equipment prices, WCR survey variables, national savings rates
DeLong and Summers (1991)
Investment – education
Age demographics (16) and lagged capital
Cook (2002b)
Investment – education
Age demographic variables
Higgins (1998)
Investment – education
Average level
Beaudry, Collard and Green (2002)
Investment
Initial values of investment/GDP, population growth and GDP Lagged investment, lagged output, lagged inflation, trade/GDP and gov spend/GDP Initial investment in sub-period, average savings rate in sub-period
Cho (1996)
Investment
Investment
Population growth
Population growth Neoclassical convergence RHS variables
Initial values of investment/GDP, population growth and GDP Average population growth over sub-period Civilian fatalities as % of population (and squared), number of months of occupation by German forces, and number of months of land battles in country
Bond, Leblebicioglu and Schiantarelli (2004)
Beaudry, Collard and Green (2002)
Cho (1996)
Beaudry, Collard and Green (2002) Cook (2002a)
Ch. 8: Growth Econometrics
661
Appendix D: Instrumental variables for non-Solow growth determinants
Variable
Instrument
Study
Capital market imperfections Capital controls Capital controls Corruption
Degree of insider trading Lagged values Lagged values Ethnolinguistic fractionalization All variables (some lagged) Initial levels of investment, openness, military expenditure and GDP per capita Various Lagged values
Bekaert, Harvey and Lundblad (2001) McKenzie (2001) Grilli and Milesi-Ferretii (1995) Mauro (1995)
Lagged values
Lundström (2002)
Religion and civil liberty measures Change in total fertility rate, educational spend/GDP, initial fertility level Kyriacou schooling data Legal origin, resource endowments, religious composition, ethnic diversity, and others
Dollar and Gatti (1999)
Legal origins and initial income Legal origin Initial values of same “Legal origin” and lagged versions of all explanatory variables Consumption, GDP, and others Lagged versions of all explanatory variables Wide variety of initial values of regressors and initial inflation Initial values of inflation and financial depth
Demetriades and Law (2004)
Coups Defense variables
Democracy Demography – Urban concentration Economic freedom Education Male and female level Changes in attainment and female/male ratio Change and level Enterprise size
Finance Development Competition Various indicators Depth
Depth Depth Various “factors”
Depth
Londregan and Poole (1990) Guaresma and Reitschuler (2003)
Tavares and Wacziarg (2001) Henderson (2000)
Klasen (2002)
Krueger and Lindahl (2001) Beck, Demirguc-Kunt and Levine (2003)
Claessens and Laeven (2003) King and Levine (1993) Levine, Loayza and Beck (2000)
Levine and Zervos (1998) Loayza and Ranciere (2002) Rousseau and Sylla (2001)
Rousseau and Wachtel (2002)
662
S.N. Durlauf et al.
Variable
Instrument
Study
Gini coefficient
Number of municipal townships in 1962, share of labor force in manufacturing in 1990, percentage of revenue from intergovernmental transfers in 1962 Lagged government change and variable reflecting composition change in the executive without a government change Various
Alesina and La Ferrara (2002)
Government change
Government expenditure and taxation Health Change in malaria infection rate Expenditure
Inflation Inflation Infrastructure Institutions Various Various
Various
Various and trust Quality Manufacturing exports Religiosity
Social infrastructure Social infrastructure
Stock markets
Six variables for % land coverage of type of forest and desert Physicians, visits, dialysis, insurance coverage, alcohol, over 65, beds Lagged explanatory variables Initial values of inflation and financial depth Lagged values Settler mortality rate Historically determined component of current institutional quality Geographically determined component of trade as fraction of GDP and linguistic origins Lagged values Mortality rates and initial income Lagged values Presence of state religion, regulation of religion, indicator of religious pluralism, and others State antiquity Distance from equator, fraction speaking primary European language, fraction speaking English, Frankel and Romer’s log predicted trade share Lagged stock market activity
Alesina et al. (1996)
Agell, Ohlsson and Thoursie (2003)
Gallup, Mellinger and Sachs (2000)
Rivera and Currais (1999)
Li and Zou (2002) Rousseau and Wachtel (2002) Esfahani and Ramirez (2003) Acemoglu, Johnson and Robinson (2002) Acemoglu et al. (2003)
Alcalá and Ciccone (2004)
Keefer and Knack (1997) Demetriades and Law (2004) Calderón, Chong and Zanforlin (2001) Barro and McCleary (2003)
Bockstette, Chanda and Putterman (2003) Hall and Jones (1999)
Harris (1997)
Ch. 8: Growth Econometrics
663
Variable
Instrument
Study
Technology gap (first difference) Trade As share of GDP
Lagged (second difference)
Hultberg, Nadiri and Sickles (2003)
Geographically determined component of trade as fraction of GDP and linguistic origins Lagged values and others unreported by author Lagged values Geographically determined component of trade as fraction of GDP Lagged values
Alcalá and Ciccone (2004)
Policy indices Policy indices
Various – Log initial GDP, broad money to GDP, gov. expenditure to GDP
Edwards (1998) Amable (2000) Frankel and Romer (1996, 1999)
Rousseau (2002)
References Abramovitz, M. (1986). “Catching up, forging ahead and falling behind”. Journal of Economic History 46, 385–406. Acemoglu, D., Johnson, S., Robinson, J. (2001). “The colonial origins of comparative development: An empirical investigation”. American Economic Review 91 (5), 1369–1401. Acemoglu, D., Johnson, S., Robinson, J. (2002). “Reversal of fortune: Geography and institutions in the making of the modern world income distribution”. Quarterly Journal of Economics 117 (4), 1231–1294. Acemoglu, D., Johnson, S., Robinson, J. (2004). “Institutions as the fundamental cause of long-run growth”. National Bureau of Economic Research Working Paper No. 10481. Acemoglu, D., Zilibotti, F. (2001). “Productivity differences”. Quarterly Journal of Economics 116, 563–606. Acemoglu, D., Johnson, S., Robinson, J., Thaicharoen, Y. (2003). “Institutional causes, macroeconomic symptoms: Volatility, crises and growth”. Journal of Monetary Economics 50 (1), 49–123. Agell, J., Ohlsson, H., Thoursie, P. (2003). “Growth effects of government expenditure and taxation in rich countries: A comment”. Mimeo, Stockholm University. Agénor, P.-R. (2004). “Macroeconomic adjustment and the poor: Analytical issues and cross-country evidence”. Journal of Economic Surveys 18 (3), 351–408. Ahn, S., Schmidt, P. (1995). “Efficient estimation of models for dynamic panel data”. Journal of Econometrics 68, 5–27. Aizenman, J., Glick, R. (2003). “Military expenditure, threats and growth”. National Bureau of Economic Research Working Paper No. 9618. Akerlof, G. (1997). “Social distance and economic decisions”. Econometrica 65 (5), 1005–1027. Alcalá, F., Ciccone, A. (2004). “Trade and productivity”. Quarterly Journal of Economics 119 (2), 613–646. Alesina, A., La Ferrara, E. (2002). “Who trusts others?”. Journal of Public Economics 85, 207–235. Alesina, A., Rodrik, D. (1994). “Distributive politics and economic growth”. Quarterly Journal of Economics 109 (2), 465–490. Alesina, A., Ozler, S., Roubini, N., Swagel, P. (1996). “Political instability and economic growth”. Journal of Economic Growth 1 (2), 189–211. Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S., Wacziarg, R. (2003). “Fractionalization”. Journal of Economic Growth 8 (2), 155–194. Amable, B. (2000). “International specialisation and economic growth”. Mimeo, University of Lille II.
664
S.N. Durlauf et al.
Anderson, G. (2003). “Making inferences about the polarization, welfare, and poverty of nations: A study of 101 countries 1970–1995”. Mimeo, University of Toronto. Journal of Applied Econometrics, in press. Anderson, G. (2004). “Toward an empirical analysis of polarization”. Journal of Econometrics 122 (1), 1–26. Anderson, G. Ge, Y. (2004). “A new approach to convergence: City types and “complete” convergence of post-reform Chinese urban income distributions”. Mimeo, University of Toronto. Andrade, E., Laurini, M., Madalozzo, R., Valls Pereira, P. (2004). “Convergence clubs among Brazilian municipalities”. Economics Letters 83, 179–184. Andres, J., Lamo, A. (1995). “Dynamics of the income distribution across OECD countries”. London School of Economics, Centre for Economic Performance Discussion Paper No. 252. Anselin, L. (2001). “Spatial econometrics”. In: Baltagi, B. (Ed.), A Companion to Theoretical Econometrics. Blackwell, Oxford. Arellano, M. (2003). Panel Data Econometrics. Oxford University Press, Oxford. Arellano, M., Bond, S. (1991). “Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations”. Review of Economic Studies 58 (2), 277–297. Arellano, M., Bover, O. (1995). “Another look at the instrumental-variable estimation of error-components models”. Journal of Econometrics 68, 29–51. Azariadis, C., Drazen, A. (1990). “Threshold externalities in economic development”. Quarterly Journal of Economics 105 (2), 501–526. Azariadis, C., Stachurski, J. (2003). “A forward projection of the cross-country income distribution”. Discussion Paper No. 570, Institute of Economic Research, Kyoto University. Baltagi, B., Song, S., Koh, W. (2003). “Testing panel data regression models with spatial error correlation”. Journal of Econometrics 117 (1), 123–150. Bandyopadhyay, S. (2002). “Polarisation, stratification, and convergence clubs: Some dynamics and explanations of unequal economic growth across Indian states”. Mimeo, London School of Economics. Banerjee, A., Duflo, E. (2003). “Inequality and growth: What can the data say?”. Journal of Economic Growth 8 (3), 267–300. Barro, R. (1991). “Economic growth in a cross section of countries”. Quarterly Journal of Economics 106 (2), 407–443. Barro, R. (1996). “Democracy and growth”. Journal of Economic Growth 1 (1), 1–27. Barro, R. (1997). Determinants of Economic Growth. MIT Press, Cambridge. Barro, R., Lee, J.-W. (1994). “Sources of economic growth (with commentary)”. Carnegie-Rochester Conference Series on Public Policy 40, 1–57. Barro, R., Mankiw, N.G., Sala-i-Martin, X. (1995). “Capital mobility in neoclassical models of growth”. American Economic Review 85 (1), 103–115. Barro, R., McCleary, R. (2003). “Religion and economic growth across countries”. American Sociological Review 68 (5), 760–781. Barro, R., Sala-i-Martin, X. (1991). “Convergence across states and regions”. Brookings Papers on Economic Activity 1, 107–158. Barro, R., Sala-i-Martin, X. (1992). “Convergence”. Journal of Political Economy 100, 223–251. Barro, R., Sala-i-Martin, X. (1997). “Technological diffusion, convergence, and growth”. Journal of Economic Growth 2, 1–26. Barro, R., Sala-i-Martin, X. (2004). Economic Growth, second ed. MIT Press, Cambridge. Baumol, W. (1986). “Productivity growth, convergence, and welfare: What the long-run data show”. American Economic Review 76 (5), 1072–1085. Beaudry, P., Collard, F., Green, D. (2002). “Decomposing the twin-peaks in the world distribution of outputper-worker”. National Bureau of Economic Research Working Paper No. 9240. Beck, T., Demirguc-Kunt, A., Levine, R. (2003). “Small and medium enterprises, growth, and poverty: Cross country evidence”. World Bank Research Policy Paper 3178. Beck, T., Levine, R. (2004). “Stock market, banks, and growth: Panel evidence”. Journal of Banking and Finance 28 (3), 423–442. Bekaert, G., Harvey, C., Lundblad, C. (2001). Does financial liberalization spur growth?. National Bureau of Economic Research Working Paper No. 8245.
Ch. 8: Growth Econometrics
665
Ben-David, D. (1993). “Equalizing exchange: Trade liberalization and income convergence”. Quarterly Journal of Economics 108, 653–679. Ben-David, D. (1996). “Trade and convergence among countries”. Journal of International Economics 40 (3/4), 279–298. Bernard, A., Durlauf, S. (1995). “Convergence in international output”. Journal of Applied Econometrics 10 (2), 97–108. Bernard, A., Durlauf, S. (1996). “Interpreting tests of the convergence hypothesis”. Journal of Econometrics 71 (1–2), 161–173. Bernard, A., Jones, C. (1996). “Comparing apples to oranges: Productivity convergence and measurement across industries and countries”. American Economic Review 86 (5), 1216–1238. Berthelemy, J., Varoudakis, A. (1996). “Economic growth, convergence clubs, and the role of financial development”. Oxford Economic Papers 48, 300–328. Besley, T., Burgess, R. (2000). “Land reform, poverty reduction, and growth: Evidence from India”. Quarterly Journal of Economics 115 (2), 389–430. Besley, T., Burgess, R. (2002). “The political economy of government responsiveness: Theory and evidence from India”. Quarterly Journal of Economics 117 (4), 1415–1451. Besley, T., Burgess, R. (2004). “Can labor regulation hinder economic performance? Evidence from India”. Quarterly Journal of Economics 119 (1), 91–134. Beugelsdijk, S., van Schalk, T. (2001). “Social capital and regional economic growth”. Mimeo, Tilburg University. Bhargava, A., Jamison, D., Lau, L., Murray, C. (2001). “Modeling the effects of health on economic growth”. Journal of Health Economics 20 (3), 423–440. Bianchi, M. (1997). “Testing for convergence: Evidence from nonparametric multimodality tests”. Journal of Applied Econometrics 12 (4), 393–409. Bils, M., Klenow, P. (2000). “Does schooling cause growth?”. American Economic Review 90 (5), 1160– 1183. Binder, M., Brock, S. (2004). “A re-examination of determinants of economic growth using simultaneous equation dynamic panel data models”. Mimeo, Johannes Goethe University, Frankfurt. Binder, M., Pesaran, M.H. (1999). “Stochastic growth models and their econometric implications”. Journal of Economic Growth 4, 139–183. Binder, M., Pesaran, M.H. (2001). “Life cycle consumption under social interactions”. Journal of Economic Dynamics and Control 25 (1–2), 35–83. Blattman, C., Hwang, J. Williamson, J. (2003). “The terms of trade and economic growth in the periphery 1870–1983”. National Bureau of Economic Research Working Paper No. 9940. Bliss, C. (1999). “Galton’s fallacy and economic convergence”. Oxford Economic Papers 51, 4–14. Bliss, C. (2000). “Galton’s fallacy and economic convergence: A reply to Cannon and Duck”. Oxford Economic Papers 52, 420–422. Blomstrom, M., Lipsey, R., Zejan, M. (1996). “Is fixed investment the key to growth?”. Quarterly Journal of Economics 111 (1), 269–276. Blonigen, B., Wang, M. (2004). “Inappropriate pooling of wealthy and poor countries in empirical FDI studies”. National Bureau of Economic Research Working Paper No. 10378. Bloom, D., Canning, D., Sevilla, J. (2003). “Geography and poverty traps”. Journal of Economic Growth 8, 355–378. Bloom, D., Canning, D., Sevilla, J. (2004). “The effect of health on economic growth: A production function approach”. World Development 32 (1), 1–13. Mimeo, Harvard University. Bloom, D., Malaney, P. (1998). “Macroeconomic consequences of the Russian mortality crisis”. World Development 26 (11), 2073–2085. Bloom, D., Sachs, J. (1998). “Geography, demography, and economic growth in Africa”. Mimeo, Harvard Institute for International Development. Bloom, D., Williamson, J. (1998). “Demographic transitions and economic miracles in emerging Asia”. World Bank Economic Review 12 (3), 419–455.
666
S.N. Durlauf et al.
Blundell, R., Bond, S. (1998). “Initial conditions and moment restrictions in dynamic panel data models”. Journal of Econometrics 87 (1), 115–143. Bockstette, V., Chanda, A., Putterman, L. (2003). “States and markets: The advantages of an early start”. Journal of Economic Growth 7, 347–369. Bond, S. (2002). “Dynamic panel data models: A guide to micro data methods and practice”. Portugese Economic Journal 1, 141–162. Bond, S., Hoeffler, A., Temple, J. (2001). “GMM estimation of empirical growth models”. Centre for Economic Policy Research Discussion Paper No. 3048. Bond, S., Leblebicioglu, A., Schiantarelli, F. (2004). “Capital accumulation and growth: A new look at the empirical evidence”. Nuffield College, Oxford, Working Paper No. 2004-W8. Bosworth, B., Collins, S. (2003). “The empirics of growth: An update (with discussion)”. Brookings Papers on Economic Activity 2, 113–206. Breiman, L., Friedman, J., Olshen, R., Stone, C. (1984). Classification and Regression Trees. Wadsworth Publishing, Redwood City. Breusch, T., Pagan, A. (1980). “The Lagrange multiplier test and its application to model specifications in econometrics”. Review of Economic Studies 47, 239–253. Brock, W., Durlauf, S. (2001a). “Growth empirics and reality”. World Bank Economic Review 15 (2), 229– 272. Brock, W., Durlauf, S. (2001b). “Interactions-based models”. In: Heckman, J., Leamer, E. (Eds.), Handbook of Econometrics, vol. 5. North-Holland, Amsterdam. Brock, W., Durlauf, S., West, K. (2003). “Policy evaluation in uncertain economic environments (with discussion)”. Brookings Papers on Economic Activity 1, 235–322. Bruno, M., Easterly, W. (1998). “Inflation crises and long-run growth”. Journal of Monetary Economics 41 (1), 3–26. Bulli, S. (2001). “Distribution dynamics and cross-country convergence: New evidence”. Scottish Journal of Political Economy 48, 226–243. Bun, M., Kiviet, J. (2001). “The accuracy of inference in small samples of dynamic panel data models”. Tinbergen Institute Discussion Paper No. 2001-006/4. Calderón, C., Chong, A., Zanforlin, L. (2001). “On the non-linearities between exports of manufactures and economic growth”. Journal of Applied Economics 4, 279–311. Campos, N., Nugent, J. (2002). “Who is afraid of political instability?”. Journal of Development Economics 67, 157–172. Cannon, E., Duck, N. (2000). “Galton’s fallacy and economic convergence”. Oxford Economic Papers 53, 415–419. Canova, F. (2004). “Testing for convergence clubs in income per capita: A predictive density approach”. International Economic Review 45 (1), 49–77. Canova, F., Marcet, A. (1995). “The poor stay poor: Non-convergence across countries and regions”. Centre for Economic Policy Research Discussion Paper 1265. Carlino, G., Mills, L. (1993). “Are U.S. regional incomes converging?: A time series analysis”. Journal of Monetary Economics 32 (2), 335–346. Caselli, F., Coleman, W.J. (2003). “The world technology frontier”. Mimeo, Harvard University. Caselli, F., Esquivel, G., Lefort, F. (1996). “Reopening the convergence debate: a new look at cross country growth empirics”. Journal of Economic Growth 1 (3), 363–389. Cashin, P. (1995). “Economic growth and convergence across the seven colonies of Australasia: 1861–1991”. Economic Record 71, 132–144. Cashin, P., Sahay, R. (1996). “Regional economic growth and convergence in India”. Finance and Development 33, 49–52. Chesher, A. (1984). “Testing for neglected heterogeneity”. Econometrica 52 (4), 865–872. Cho, D. (1996). “An alternative interpretation of conditional convergence results”. Journal of Money, Credit and Banking 28 (1), 669–681. Ciccone, A. (1996). “Externalities and interdependent growth: Theory and evidence”. Mimeo, UC-Berkeley.
Ch. 8: Growth Econometrics
667
Claessens, S., Laeven, L. (2003). “Competition in the financial sector and growth: A cross country perspective”. Mimeo, University of Amsterdam. Cohen, D. (1996). “Tests of the convergence hypothesis: Some further results”. Journal of Economic Growth 1 (3), 351–361. Collier, P., Gunning, J. (1999a). “Why has Africa grown slowly?”. Journal of Economic Perspectives 13 (3), 3–22. Collier, P., Gunning, J. (1999b). “Explaining African economic performance”. Journal of Economic Literature 37 (1), 64–111. Conley, T. (1999). “GMM estimation with cross-section dependence”. Journal of Econometrics 92, 1–45. Conley, T., Ligon, E. (2002). “Economic distance and long-run growth”. Journal of Economic Growth 7 (2), 157–187. Conley, T., Topa, G. (2002). “Socio-economic distance and spatial patterns in unemployment”. Journal of Applied Econometrics 17 (4), 303–327. Cook, D. (2002a). “World War II and convergence”. Review of Economics and Statistics 84 (1), 131–138. Cook, D. (2002b). “Education and growth: Instrumental variables estimates”. Mimeo, Hong Kong University of Science and Technology. Corrado, L., Martin, R., Weeks, M. (2004). “Identifying and interpreting regional convergence clusters across Europe”. Economic Journal, in press. Coulombe, S., Lee, F. (1995). “Convergence across Canadian provinces, 1961 to 1991”. Canadian Journal of Economics 28, 886–898. Dagenais, M., Dagenais, D. (1997). “Higher moment estimators for linear regression models with errors in the variables”. Journal of Econometrics 76 (1–2), 193–221. Davidson, R., MacKinnon, J. (1993). Estimation and Inference in Econometrics. Oxford University Press, Oxford. De Gregorio, J., Guidotti, P. (1995). “Financial development and economic growth”. World Development 23 (3), 433–448. Deidda, L., Fattouh, B. (2002). “Non-linearity between finance and growth”. Economics Letters 74, 339–345. DeLong, J.B. (1988). “Productivity growth, convergence, and welfare: Comment”. American Economic Review 78 (5), 1138–1154. DeLong, J.B., Summers, L. (1991). “Equipment investment and economic growth”. Quarterly Journal of Economics 106 (2), 445–502. DeLong, J.B., Summers, L. (1993). “How strongly do developing economies benefit from equipment investment?”. Journal of Monetary Economics 32 (3), 395–415. Demetriades, P., Law, S. (2004). “Finance, institutions and economic growth”. University of Leicester Working Paper 04/5. Desdoigts, A. (1999). “Patterns of economic development and the formation of clubs”. Journal of Economic Growth 4 (3), 305–330. Diebold, F., Inoue, A. (2001). “Long memory and regime switching”. Journal of Econometrics 105 (1), 131– 159. Dollar, D. (1992). “Outward-oriented developing economies really do grow more rapidly: Evidence from 95 LDCs, 1976–85”. Economic Development and Cultural Change 40, 523–544. Dollar, D., Gatti, R. (1999). “Gender inequality, income, and growth: Are good times good for women?”. Mimeo, World Bank. Dollar, D., Kraay, A. (2003). “Institutions, trade and growth: Revisiting the evidence”. Journal of Monetary Economics 50 (1), 133–162. Doppelhofer, G., Miller, R., Sala-i-Martin, X. (2004). “Determinants of long-term growth: A Bayesian Averaging of Classical Estimates (BACE) approach”. American Economic Review 94 (4), 813–835. Dowrick, S. (2004). “De-linearising the neo-classical convergence model”. In: Turnovsky, S., Dowrick, S., Pitchford, R. (Eds.), Economic Growth and Macrodynamics: Recent Developments in Economic Theory. Cambridge University Press, New York. Dowrick, S., Quiggin, J. (1997). “True measures of GDP and convergence”. American Economic Review 87 (1), 41–64.
668
S.N. Durlauf et al.
Dowrick, S., Rogers, M. (2002). “Classical and technological convergence: Beyond the Solow–Swan growth model”. Oxford Economic Papers 54, 369–385. Draper, D. (1995). “Assessment and propagation of model uncertainty”. Journal of the Royal Statistical Society, Series B 57, 45–70. Draper, D., Hodges, J., Mallows, C., Pregibon, D. (1993). “Exchangeability and data analysis (with discussion)”. Journal of the Royal Statistical Society, Series A 156, 9–37. Driscoll, J., Kraay, A. (1998). “Consistent covariance matrix estimation with spatially dependent panel data”. Review of Economics and Statistics 80 (4), 549–560. Duffy, J., Papageorgiou, C. (2000). “A cross-country empirical investigation of the aggregate production function specification”. Journal of Economic Growth 5, 87–120. Durlauf, S. (1996). “Controversy on the convergence and divergence of growth rates”. Economic Journal 106, 1016–1018. Durlauf, S. (2001). “Manifesto for a growth econometrics”. Journal of Econometrics 100 (1), 65–69. Durlauf, S. (2002). “On the empirics of social capital”. Economic Journal 112, 459–479. Durlauf, S., Johnson, P. (1995). “Multiple regimes and cross country growth behaviour”. Journal of Applied Econometrics 10 (4), 365–384. Durlauf, S., Kourtellos, A., Minkin, A. (2001). “The local Solow growth model”. European Economic Review 45 (4–6), 928–940. Durlauf, S., Quah, D. (1999). “The new empirics of economic growth”. In: Taylor, J., Woodford, M. (Eds.), Handbook of Macroeconomics. North-Holland, Amsterdam. Easterly, W. (1993). “How much do distortions affect growth?”. Journal of Monetary Economics 32 (2), 187–212. Easterly, W. (1994). “Economic stagnation, fixed factors, and policy thresholds”. Journal of Monetary Economics 33 (3), 525–557. Easterly, W. (1996). “When is stabilization expansionary?”. Economic Policy 22, 67–98. Easterly, W. (2001). “The lost decades: Developing countries’ stagnation in spite of policy reform 1980– 1998”. Journal of Economic Growth 6 (2), 135–157. Easterly, W., Levine, R. (1997a). “Africa’s growth tragedy: Policies and ethnic divisions”. Quarterly Journal of Economics 112 (4), 1203–1250. Easterly, W., Levine, R. (1997b). “Troubles with the neighbours: Africa’s problem, Africa’s opportunity”. Journal of African Economies 7 (1), 120–142. Easterly, W., Levine, R. (2001). “It’s not factor accumulation: Stylized facts and growth models”. World Bank Economic Review 15, 177–219. Easterly, W., Kremer, M., Pritchett, L., Summers, L. (1993). “Good policy or good luck? Country growth performance and temporary shocks”. Journal of Monetary Economics 32, 459–483. Eaton, J., Kortum, S. (1999). “International technology diffusion: Theory and measurement”. International Economic Review 40 (3), 537–570. Eaton, J., Kortum, S. (2001). “Trade in capital goods”. European Economic Review 4 (7), 1195–1235. Edwards, S. (1998). “Openness, productivity and growth: What do we really know?”. Economic Journal 108, 383–398. Edwards, S., Magendzo, I. (2003). “Strict dollarization and economic performance: An empirical investigation”. International Journal of Finance and Economics 8 (4), 351–363. Eichengreen, B., Leblang, D. (2003). “Capital account liberalization and growth: Was Mr. Mahathir right?”. International Journal of Finance and Economics 8 (3), 205–224. Eicker, F. (1967). “Limit theorems for regressions with unequal and dependent errors”. In: Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability, vol. 1. University of California, Berkeley. Esfahani, H., Ramirez, M. (2003). “Institutions, infrastructure, and economic growth”. Journal of Development Economics 70 (2), 443–477. Evans, P. (1996). “Using cross-country variances to evaluate growth theories”. Journal of Economic Dynamics and Control 20, 1027–1049.
Ch. 8: Growth Econometrics
669
Evans, P. (1997). “How fast do economies converge?”. Review of Economics and Statistics 79 (2), 219–225. Evans, P. (1998). “Using panel data to evaluate growth theories”. International Economic Review 39 (2), 295–306. Feder, G. (1982). “On exports and economic growth”. Journal of Development Economics 12 (1), 59–74. Feld, L., Voigt, S. (2003). “Economic growth and judicial independence: Cross country evidence using a new set of indicators”. European Journal of Political Economy 19 (3), 497–527. Fernandez, C., Ley, E., Steel, M. (2001a). “Model uncertainty in cross-country growth regressions”. Journal of Applied Econometrics 16 (5), 563–576. Fernandez, C., Ley, E., Steel, M. (2001b). “Benchmark priors for Bayesian model averaging”. Journal of Econometrics 100 (2), 381–427. Feyrer, J. (2003). “Convergence by parts”. Mimeo, Dartmouth College. Fiaschi, D., Lavezzi, A. (2004). “Distribution dynamics and nonlinear growth”. Journal of Economic Growth 8, 379–401. Fischer, S. (1993). “The role of macroeconomic factors in growth”. Journal of Monetary Economics 32 (3), 485–512. Forbes, K. (2000). “A reassessment of the relationship between inequality and growth”. American Economic Review 90 (4), 869–887. Frankel, J. (2003). “Discussion”. Brookings Papers on Economic Activity 2, 189–199. Frankel, J., Romer, D. (1996). “Trade and growth: An empirical investigation”. National Bureau of Economic Research Working Paper No. 5476. Frankel, J., Romer, D. (1999). “Does trade cause growth?”. American Economic Review 89 (3), 379–399. Friedman, J. (1987). “Exploratory projection pursuit”. Journal of the American Statistical Association 82, 249–266. Friedman, J., Tukey, J. (1974). “A projection pursuit algorithm for exploratory data analysis”. IEEE Transactions on Computers C23, 881–890. Friedman, M. (1992). “Do old fallacies ever die?”. Journal of Economic Literature 30, 2129–2132. Galor, O. (1996). “Convergence? Inferences from theoretical models”. Economic Journal 106, 1056–1069. Gallup, J., Mellinger, A., Sachs, J. (2000). “The economic burden of malaria”. Harvard University, Center for International Development Working Paper No. 1. Giavazzi, F. Tabellini, G. (2004). “Economic and political liberalizations”. Centre for Economic Policy Research Discussion Paper No. 4579. Goetz, S., Hu, D. (1996). “Economic growth and human capital accumulation: Simultaneity and expanded convergence tests”. Economic Letters 51, 355–362. Graham, B., Temple, J. (2003). “Rich nations, poor nations: How much can multiple equilibria explain?”. Mimeo, University of Bristol. Granato, J., Inglehart, R., Leblang, D. (1996). “The effect of cultural values on economic development: theory, hypotheses, and some empirical tests”. American Journal of Political Science 40 (3), 607–631. Granger, C.W.J. (1980). “Long-memory relationships and the aggregation of dynamic models”. Journal of Econometrics 14, 227–238. Greasley, D., Oxley, L. (1997). “Time-series tests of the convergence hypothesis: Some positive results”. Economics Letters 56, 143–147. Grier, K., Tullock, G. (1989). “An empirical analysis of cross national economic growth, 1951–80”. Journal of Monetary Economics 24 (2), 259–276. Grilli, V., Milesi-Ferretii, G. (1995). “Economic effects and structural determinants of capital controls”. International Monetary Fund Working Paper WP/95/31. Guaresma, J., Reitschuler, G. (2003). “‘Guns or butter?’ revisited: Robustness and nonlinearity issues in the defense-growth nexus”. Mimeo, University of Vienna. Hahn, J. (1999). “How informative is the initial condition in the dynamic panel model with fixed effects?”. Journal of Econometrics 93 (2), 309–326. Hahn, J., Hausman, J., Kuersteiner, G. (2001). “Bias corrected instrumental variables estimation for dynamic panel models with fixed effects”. Mimeo, Massachusetts Institute for Technology.
670
S.N. Durlauf et al.
Hall, A. (1987). “The information matrix test for the linear model”. Review of Economic Studies 54 (2), 257–263. Hall, R., Jones, C. (1999). “Why do some countries produce so much more output per worker than others?”. Quarterly Journal of Economics 114 (1), 83–116. Hall, S., Robertson, D., Wickens, M. (1997). “Measuring economic convergence”. International Journal of Finance and Economics 2, 131–143. Hamoudi, A., Sachs, J. (2000). “Economic consequences of health status: A review of the evidence”. Harvard University, CID Working Paper No. 30. Hansen, B. (2000). “Sample splitting and threshold estimation”. Econometrica 68 (3), 575–603. Hanushek, E., Kimko, D. (2000). “Schooling, labor-force quality, and the growth of nations”. American Economic Review 90 (5), 1184–1208. Harberger, A. (1987). “Comment”. In: Fischer, S. (Ed.), Macroeconomics Annual 1987. MIT Press, Cambridge. Harris, D. (1997). “Stock markets and development: a re-assessment”. European Economic Review 41 (1), 139–146. Harrison, A. (1996). “Openness and growth: A time-series, cross-country analysis for developing countries”. Journal of Development Economics 48 (2), 419–447. Hausmann, R., Pritchett, L., Rodrik, D. (2004). “Growth accelerations”. Centre for Economic Policy Research Discussion Paper No. 4538. Helliwell, J. (1996). “Economic growth and social capital in Asia”. In: Harris, R. (Ed.), The Asia Pacific Region in the Global Economy: A Canadian Perspective. University of Calgary Press, Calgary. Helliwell, J., Putnam, R. (2000). “Economic growth and social capital in Italy”. In: Dasgupta, P., Seragilden, I. (Eds.), Social Capital: A Multifaceted Perspective. World Bank, Washington, DC. Henderson, D., Russell, R. (2004). “Human capital and convergence: A production frontier approach”. Mimeo, SUNY Binghamton. International Economic Review, in press. Henderson, J.V. (2000). “The effects of urban concentration on economic growth”. National Bureau of Economic Research Working Paper No. 7503. Hendry, D. (1995). Dynamic Econometrics. Oxford University Press, New York. Hendry, D., Krolzig, H.-M. (2004). “We ran one regression”. Mimeo, Oxford University. Hendry, D., Krolzig, H.-M. (2005). “The properties of automatic gets modelling”. Economic Journal, in press. Henry, P. (2000). “Do stock market liberalizations cause investment booms?”. Journal of Financial Economics 58 (1–2), 301–334. Henry, P. (2003). “Capital account liberalization, the cost of capital, and economic growth”. American Economic Review 93 (2), 91–96. Heston, A., Summers, R., Aten, B. (2002). “Penn World Table version 6.1”. Center for International Comparisons at the University of Pennsylvania (CICUP). Higgins, M. (1998). “Demography, national savings and international capital flows”. International Economic Review 39, 343–369. Hobijn, B., Franses, P. (2000). “Asymptotically perfect and relative convergence of productivity”. Journal of Applied Econometrics 15, 59–81. Hoeffler, A. (2002). “The augmented Solow model and the African growth debate”. Oxford Bulletin of Economics and Statistics 64 (2), 135–158. Holtz-Eakin, D., Newey, W., Rosen, H. (1988). “Estimating vector autoregressions with panel data”. Econometrica 56 (6), 1371–1395. Hoover, K., Perez, S. (2004). “Truth and robustness in cross-country growth regressions”. Oxford Bulletin of Economics and Statistics, in press. Howitt, P. (2000). “Endogenous growth and cross-country income differences”. American Economic Review 90 (4), 829–846. Hultberg, P., Nadiri, M., Sickles, R. (2003). “Cross-country catch-up in the manufacturing sector: Impacts of heterogeneity on convergence and technology adoption”. Mimeo, University of Wyoming. Hulten, C. (1996). “Infrastructure capital and economic growth: How well you use it may be more important than how much you have”. National Bureau of Economic Research Working Paper No. 5847.
Ch. 8: Growth Econometrics
671
Islam, N. (1995). “Growth empirics: A panel data approach”. Quarterly Journal of Economics 110 (4), 1127– 1170. Islam, N. (1998). “Growth empirics: A panel data approach – a reply”. Quarterly Journal of Economics 113, 325–329. Islam, N. (2003). “What have we learned from the convergence debate?”. Journal of Economic Surveys 17, 309–362. Johnson, P. (2000). “A nonparametric analysis of income convergence across the US States”. Economics Letters 69, 219–223. Johnson, P. (2004). “A continuous state space approach to ‘convergence by parts’ ”. Mimeo, Vassar College. Economic Letters, in press. Johnson, P., Takeyama, L. (2001). “Initial conditions and economic growth in the US states”. European Economic Review 45 (4–6), 919–927. Jones, C. (1995). “Time series tests of endogenous growth models”. Quarterly Journal of Economics 110 (2), 495–525. Jones, C. (1997). “Convergence revisited”. Journal of Economic Growth 2 (2), 131–153. Jones, L., Manuelli, R. (1990). “A convex model of equilibrium growth: Theory and policy implications”. Journal of Political Economy 98 (5), 1008–1038. Judson, R., Owen, A. (1999). “Estimating dynamic panel data models: A guide for macroeconomists”. Economics Letters 65, 9–15. Kalaitzidakis, P., Mamuneas, T., Stengos, T. (2000). “A non-linear sensitivity analysis of cross country growth regressions”. Canadian Journal of Economics 33 (3), 604–617. Kaufmann, D., Kraay, A., Mastruzzi, M. (2003). “Governance matters III: Governance indicators for 1996– 2002”. Mimeo, World Bank. Kaufmann, D., Kraay, A., Zoido-Lobaton, P. (1999a). “Aggregating governance indicators”. World Bank Policy Research Department Working Paper No. 2195. Kaufmann, D., Kraay, A., Zoido-Lobaton, P. (1999b). “Governance matters”. World Bank Policy Research Department Working Paper No. 2196. Keefer, P., Knack, S. (1997). “Why don’t poor countries catch up? A cross-national test of an institutional explanation”. Economic Inquiry 35 (3), 590–602. Kelley, A., Schmidt, R. (1995). “Aggregate population and economic growth correlations: The role of the components of demographic change”. Demography 32, 543–555. Kelly, M. (1992). “On endogenous growth with productivity shocks”. Journal of Monetary Economics 30 (1), 47–56. Kelly, T. (1997). “Public expenditures and growth”. Journal of Development Studies 34 (1), 60–84. King, R., Levine, R. (1993). “Finance and growth: Schumpeter might be right”. Quarterly Journal of Economics 108 (3), 717–737. Kiviet, J. (1995). “On bias, inconsistency, and efficiency of various estimators in dynamic panel data models”. Journal of Econometrics 68 (1), 53–78. Kiviet, J. (1999). “Expectations of expansions for estimators in a dynamic panel data model: Some results for weakly exogenous regressors”. In: Hsiao, C. (Ed.), Analysis of Panels and Limited Dependent Variable Models: In Honor of G.S. Maddala. Cambridge University Press, Cambridge. Klasen, S. (2002). “Lower schooling for girls, slower growth for all? Cross-country evidence on the effect of gender inequality in education on economic development”. World Bank Economic Review 16 (3), 345–373. Klenow, P., Rodriguez-Clare, A. (1997a). “The neoclassical revival in growth economics: has it gone too far?”. In: Bernanke, B., Rotemberg, J. (Eds.), Macroeconomics Annual 1997. MIT Press, Cambridge. Klenow, P., Rodriguez-Clare, A. (1997b). “Economic growth: a review essay”. Journal of Monetary Economics 40, 597–617. Klepper, S. (1988). “Regression diagnostics for the classical errors-in-variables model”. Journal of Econometrics 37, 225–250. Klepper, S., Leamer, E. (1984). “Consistent sets of estimates for regressions with errors in all variables”. Econometrica 52 (1), 163–183.
672
S.N. Durlauf et al.
Knack, S. (1999). “Social capital, growth and poverty: a survey of cross country evidence”. World Bank, Social Capital Initiative Working Paper No. 7. Knack, S., Keefer, P. (1995). “Institutions and economic performance: Cross country tests using alternative institutional measures”. Economics and Politics 7 (3), 207–227. Knack, S., Keefer, P. (1997). “Does social capital have an economic payoff? A cross-country investigation”. Quarterly Journal of Economics 112 (4), 1252–1288. Knight, M., Loayza, N., Villanueva, D. (1993). “Testing the neoclassical growth model”. IMF Staff Papers 40, 512–541. Knowles, S. (2001). “Inequality and economic growth: The empirical relationship reconsidered in the light of comparable data”, CREDIT Research Paper 01/03. Knowles, S., Owen, P. (1995). “Health capital and cross-country variation in income per capita in the Mankiw–Romer–Weil model”. Economics Letters 48 (1), 99–106. Kocherlakota, N., Yi, K.-M. (1997). “Is there endogenous long-run growth? Evidence from the United States and the United Kingdom”. Journal of Money, Credit and Banking 29 (2), 235–262. Kormendi, R., Meguire, P. (1985). “Macroeconomic determinants of growth: Cross country evidence”. Journal of Monetary Economics 16 (2), 141–163. Kourtellos, A. (2003a). “Modeling parameter heterogeneity in cross-country growth regression models”. Mimeo, University of Cyprus. Kourtellos, A. (2003b). “A projection pursuit approach to cross-country growth data”. Mimeo, University of Cyprus. Krasker, W., Pratt, J. (1986). “Bounding the effects of proxy variables on regression coefficients”. Econometrica 54 (3), 641–655. Krasker, W., Pratt, J. (1987). “Bounding the effects of proxy variables on instrumental variables coefficients”. Journal of Econometrics 35 (2/3), 233–252. Kremer, M., Onatski, A., Stock, J. (2001). “Searching for prosperity”. Carnegie-Rochester Conference Series on Public Policy 55, 275–303. Krueger, A., Lindahl, M. (2001). “Education for growth: Why and for whom”. Journal of Economic Literature 39 (4), 1101–1136. Lamo, A. (2000). “On convergence empirics: some evidence for Spanish regions”. Investigaciones Economicas 24, 681–707. Landes, D. (1998). The Wealth and Poverty of Nations: Why Some Are so Rich and Some so Poor. W.W. Norton and Company, New York. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R. (1997). “Trust in large organizations”. American Economic Review 87 (2), 333–338. Leamer, E. (1978). Specification Searches. Wiley, New York. Leamer, E. (1983). “Let’s take the con out of econometrics”. American Economic Review 73 (1), 31–43. Leamer, E., Leonard, H. (1983). “Reporting the fragility of regression estimates”. Review of Economics and Statistics 65 (2), 306–317. Lee, K., Pesaran, M., Smith, R. (1997). “Growth and convergence in multi country empirical stochastic Solow model”. Journal of Applied Econometrics 12 (4), 357–392. Lee, K., Pesaran, M., Smith, R. (1998). “Growth empirics: a panel data approach: A comment”. Quarterly Journal of Economics 113 (1), 319–323. Levine, R., Loayza, N., Beck, T. (2000). “Financial intermediation and growth: causality and causes”. Journal of Monetary Economics 46 (1), 31–77. Levine, R., Renelt, D. (1991). “Cross-country studies of growth and policy: Methodological, conceptual, and statistical problems”. World Bank PRE Working Paper No. 608. Levine, R., Renelt, D. (1992). “A sensitivity analysis of cross-country growth regressions”. American Economic Review 82 (4), 942–963. Levine, R., Zervos, S. (1993). “What we have learned about policy and growth from cross-country regressions”. American Economic Review 83 (2), 426–430. Levine, R., Zervos, S. (1998). “Stock markets, banks and economic growth”. American Economic Review 88 (3), 537–558.
Ch. 8: Growth Econometrics
673
Li, H., Zou, H.-F. (2002). “Inflation, growth, and income distribution: A cross country study”. Annals of Economics and Finance 3 (1), 85–101. Li, Q., Papell, D. (1999). “Convergence of international output: Time series evidence for 16 countries”. International Review of Economics and Finance 8, 267–280. Lichtenberg, F. (1992). “R&D Investment and international productivity differences”. National Bureau of Economic Research Working Paper No. 4161. Liu, Z., Stengos, T. (1999). “Non-linearities in cross country growth regressions: A semiparametric approach”. Journal of Applied Econometrics 14 (5), 527–538. Loayza, N., Ranciere, R., “Financial development, financial fragility, and growth”. CESifo Working Paper Series No. 684. Loewy, M., Papell, D. (1996). “Are U.S. regional incomes converging? Some further evidence”. Journal of Monetary Economics 38 (3), 587–598. Loh, W.-Y. (2002). “Regression trees with unbiased variable selection and interaction detection”. Statistica Sinica 12, 361–386. Londregan, J., Poole, K. (1990). “Poverty, the coup trap, and the seizure of executive power”. World Politics 4 (2), 151–183. Long, J., Ervin, L. (2000). “Using heteroscedasticity consistent standard errors in the linear regression model”. American Statistician 54 (3), 217–224. Lucas, R. (1988). “On the mechanics of economic development”. Journal of Monetary Economics 22 (1), 3–42. Lundström, S. (2002). “On institutions, economic growth, and the environment”. Mimeo, Göteborg University. Maasoumi, E., Racine, J., Stengos, T. (2003). “Growth and convergence: A profile of distribution dynamics and mobility”, Mimeo, Southern Methodist University. MacKinnon, J., White, H. (1985). “Some heteroskedasticity-consistent covariance matrix estimators with improved finite sample properties”. Journal of Econometrics 29, 305–325. Maddala, G., Wu, S. (2000). “Cross-country growth regressions: Problems of heterogeneity, stability, and interpretation”. Applied Economics 32, 635–642. Maddison, A. (1982). Phases of Capitalist Development. Oxford University Press, New York. Maddison, A. (1989). The World Economy in the 20th Century. OECD, Paris. Mamuneas, T., Savvides, A., Stengos, T. (2004). “Economic development and return to human capital: a smooth coefficient semiparametric approach”. Mimeo, University of Guelph. Journal of Applied Econometrics, in press. Mankiw, N.G. (1995). “The growth of nations”. Brookings Papers on Economic Activity 1, 275–310. Mankiw, N.G., Romer, D., Weil, D.N. (1992). “A contribution to the empirics of economic growth”. Quarterly Journal of Economics 107 (2), 407–437. Manski, C. (1993). “Identification of endogenous social effects: The reflection problem”. Review of Economic Studies 60 (3), 531–542. Masanjala, W., Papageorgiou, C. (2004). “Rough and lonely road to prosperity: a reexamination of the sources of growth in Africa using Bayesian model averaging”. Mimeo, Louisiana State University. Masters, W., McMillan, M. (2001). “Climate and scale in economic growth”. Journal of Economic Growth 6 (3), 167–186. Masters, W., Sachs, J. (2001). “Climate and development”. Mimeo, Purdue University. Mauro, P. (1995). “Corruption and growth”. Quarterly Journal of Economics 110 (3), 681–713. McArthur, J., Sachs, J. (2001). “Institutions and geography: Comment on Acemoglu, Johnson and Robinson”. National Bureau of Economic Research Working Paper No. 8114. McCarthy, D., Wolf, H., Wu, Y. (2000). “The growth costs of malaria”. National Bureau of Economic Research Working Paper No. 7541. McKenzie, D. (2001). “The impact of capital controls on growth convergence”. Journal of Economic Development 26 (1), 1–24. Michelacci, C., Zaffaroni, P. (2000). “Fractional (beta) convergence”. Journal of Monetary Economics 45, 129–153.
674
S.N. Durlauf et al.
Miguel, E., Satyanath, S., Segenti, E. (2003). “Economic shocks and civil conflict: an instrumental variables approach”. Mimeo, UC Berkeley. Minier, J. (1998). “Democracy and growth: alternative approaches”. Journal of Economic Growth 3 (3), 241– 266. Mokyr, J. (1992). Lever of Riches: Technological Creativity and Economic Progress. Princeton University Press, Princeton. Motley, B. (1998). “Growth and inflation: A cross-country study”. FRBSF Economic Review 1. Murphy, K., Shleifer, A., Vishny, R. (1991). “The allocation of talent: Implications for growth”. Quarterly Journal of Economics 106 (2), 503–530. Nelson, M., Singh, R. (1994). “The deficit-growth connection: Some recent evidence from developing countries”. Economic Development and Cultural Change 42, 167–191. Nerlove, M. (1999). “Properties of alternative estimators of dynamic panel models: An empirical analysis of cross-country data for the study of economic growth”. In: Hsiao, C. (Ed.), Analysis of Panels and Limited Dependent Variable Models: In Honor of G.S. Maddala. Cambridge University Press, Cambridge. Nerlove, M. (2000). “Growth rate convergence, fact or artifact? An essay on panel data econometrics”. In: Ronchetti, E. (Ed.), Panel Data Econometrics: Future Directions: Papers in Honor of Professor Pietro Balestra. North-Holland, Amsterdam. Nickell, S. (1981). “Biases in dynamic models with fixed effects”. Econometrica 49 (6), 1417–1426. Odedokun, M. (1996). “Alternative econometric approaches for analysing the role of the financial sector in economic growth: time series evidence from LDCs”. Journal of Development Economics 50 (1), 119–146. Olson, M. (1982). The Rise and Decline of Nations. Yale University Press, New Haven. Paap, R., van Dijk, H. (1998). “Distribution and mobility of wealth of nations”. European Economic Review 42 (7), 1269–1293. Papageorgiou, C. (2002). “Trade as a threshold variable for multiple regimes”. Economics Letters 71 (1), 85–91. Papageorgiou, C., Masanjala, W. (2004). “The Solow model with CES technology: Nonlinearities with parameter heterogeneity”. Journal of Applied Econometrics 19 (2), 171–201. Perron, P. (1989). “The great crash, the oil price shock, and the unit root hypothesis”. Econometrica 57 (6), 1361–1401. Persson, J. (1997). “Convergence across Swedish counties, 1911–1993”. European Economic Review 41, 1835–1852. Persson, T., Tabellini, G. (1994). “Is inequality harmful for growth?”. American Economic Review 84 (3), 600–621. Persson, T., Tabellini, G. (2003). The Economic Effects of Constitutions. MIT Press, Cambridge. Pesaran, M.H. (2004a). “A pair-wise approach to testing for output and growth convergence”. Mimeo, University of Cambridge. Pesaran, M.H. (2004b). “General diagnostic tests for cross-section dependence in panels”. Mimeo, University of Cambridge. Pesaran, M.H., Shin, Y., Smith, R. (1999). “Pooled mean group estimation of dynamic heterogeneous panels”. Journal of the American Statistical Association 94 (446), 621–634. Pesaran, M.H., Smith, R. (1995). “Estimating long-run relationships from dynamic heterogeneous panels”. Journal of Econometrics 68 (1), 79–113. Phillips, P., Sul, D. (2002). “Dynamic panel estimation and homogeneity testing under cross-section dependence”. Cowles Foundation Discussion Paper No. 1362. Phillips, P., Sul, D. (2003). “The elusive empirical shadow of growth convergence”. Cowles Foundation Discussion Paper No. 1398. Prescott, E. (1998). “Needed: A theory of total factor productivity”. International Economic Review 39, 525– 551. Pritchett, L. (1997). “Divergence, big time”. Journal of Economic Perspectives 11 (3), 3–17. Pritchett, L. (2000a). “Understanding patterns of economic growth: searching for hills among plateaus, mountains, and plains”. World Bank Economic Review 14 (2), 221–250.
Ch. 8: Growth Econometrics
675
Pritchett, L. (2000b). “The tyranny of concepts: CUDIE (cumulated, depreciated, investment effort) is not capital”. Journal of Economic Growth 5 (4), 361–384. Putnam, R., Leonardi, R., Nanetti, R. (1993). “Making Democracy Work”. Princeton University Press, Princeton. Quah, D. (1993a). “Galton’s fallacy and tests of the convergence hypothesis”. Scandinavian Journal of Economics 95, 427–443. Quah, D. (1993b). “Empirical cross-section dynamics in economic growth”. European Economic Review 37 (2–3), 426–434. Quah, D. (1996a). “Twin peaks: growth and convergence in models of distribution dynamics”. Economic Journal 106 (437), 1045–1055. Quah, D. (1996b). “Empirics for economic growth and convergence”. European Economic Review 40 (6), 1353–1375. Quah, D. (1996c). “Convergence empirics across economies with (some) capital mobility”. Journal of Economic Growth 1 (1), 95–124. Quah, D. (1997). “Empirics for growth and distribution: Stratification, polarization, and convergence clubs”. Journal of Economic Growth 2 (1), 27–59. Quah, D. (2001). “Searching for prosperity: A comment”. Carnegie-Rochester Conference Series on Public Policy 55, 305–319. Ram, R. (1999). “Financial development and economic growth”. The Journal of Development Studies 27 (2), 151–167. Ramey, G., Ramey, V. (1995). “Cross-country evidence on the link between volatility and growth”. American Economic Review 85 (5), 1138–1151. Reichlin, L. (1999). “Discussion of ‘convergence as distribution dynamics’, by Danny Quah”. In: Baldwin, R., Cohen, D., Sapir, A., Venables, A. (Eds.), Market Integration, Regionalism, and the Global Economy. Cambridge University Press, Cambridge. Rivera, B., Currais, L. (1999). “Economic growth and health: direct impact or reverse causation?”. Applied Economics Letters 6 (11), 761–764. Robertson, D., Symons, J. (1992). “Some strange properties of panel data estimators”. Journal of Applied Econometrics 7 (2), 175–189. Rodriguez, F., Rodrik, D. (2001). “Trade policy and economic growth: A sceptic’s guide to the cross-national evidence”. In: Bernanke, B., Rogoff, K. (Eds.), Macroeonomics Annual 2000. MIT Press, Cambridge. Rodrik, D. (1999). “Where did all the growth go? External shocks, social conflict, and growth collapses”. Journal of Economic Growth 4 (4), 385–412. Rodrik, D. (Ed.) (2003). In Search of Prosperity: Analytic Narratives on Economic Growth. Princeton University Press, Princeton. Rodrik, D., Subramanian, A., Trebbi, F. (2004). “Institutions rule: the primacy of institutions over geography and integration in economic development”. Journal of Economic Growth 9 (2), 131–165. Romer, D. (2001). Advanced Macroeconomics. McGraw-Hill, New York. Romer, P. (1986). “Increasing returns and long-run growth”. Journal of Political Economy 94 (5), 1002–1037. Romer, P. (1990). “Human capital and growth: theory and evidence”. Carnegie-Rochester Series on Public Policy 32, 251–286. Romer, P. (1993). “Idea gaps and object gaps in economic development”. Journal of Monetary Economics 32 (3), 543–573. Roubini, N., Sala-i-Martin, X. (1992). “Financial repression and economic growth”. Journal of Development Economics 39, 5–30. Rousseau, P. (2002). “Historical perspectives on financial development and economic growth”. Review Federal Reserve Bank of St. Louis 84 (4). Rousseau, P., Sylla, R. (2001). “Financial systems, economic growth, and globalization”. Mimeo, Vanderbilt University. Rousseau, P., Wachtel, P. (2002). “Inflation thresholds and the finance-growth nexus”. Journal of International Money and Finance 21 (6), 77–793.
676
S.N. Durlauf et al.
Rupasingha, A., Goetz, S., Freshwater, D. (2000). “Social capital and economic growth: A county-level analysis”. Journal of Agricultural and Applied Economics 32 (3), 565–572. Sachs, J. (2003). “Institutions don’t rule: direct effects of geography on per capita income”. National Bureau of Economic Research Working Paper No. 9490. Sachs, J., Warner, A. (1995). “Economic reform and the process of global integration (with discussion)”. Brookings Papers on Economic Activity 1, 1–118. Sachs, J., Warner, A. (1996). “Natural resource abundance and economic growth”. National Bureau of Economic Research Working Paper No. 5398. Sala-i-Martin, X. (1991). “Growth, macroeconomics, and development: comments”. In: Blanchard, O., Fischer, S. (Eds.), Macroeconomics Annual 1991. MIT Press, Cambridge. Sala-i-Martin, X. (1996a). “The classical approach to convergence analysis”. Economic Journal 106, 1019– 1036. Sala-i-Martin, X. (1996b). “Regional cohesion: evidence and theories of regional growth and convergence”. European Economic Review 40, 1325–1352. Sala-i-Martin, X. (1997a). “I just ran 4 million regressions”. National Bureau of Economic Research Working Paper No. 6252. Sala-i-Martin, X. (1997b). “I just ran 2 million regressions”. American Economic Review 87 (2), 178–183. Shioji, E. (2001a). “Composition effect of migration and regional growth in Japan”. Journal of the Japanese and International Economies 15, 29–49. Shioji, E. (2001b). “Public capital and economic growth: A convergence approach”. Journal of Economic Growth 6 (3), 205–227. Solow, R. (1956). “A contribution to the theory of economic growth”. Quarterly Journal of Economics 70 (1), 65–94. Solow, R. (1994). “Perspectives on growth theory”. Journal of Economic Perspectives 8, 45–54. Summers, R., Heston, A. (1988). “A new set of international comparisons of real product and price levels estimates for 130 countries, 1950–1985”. Review of Income and Wealth 34, 1–25. Summers, R., Heston, A. (1991). “The Penn World Table (Mark 5): An expanded set of international comparisons, 1950–1988”. Quarterly Journal of Economics 106 (2), 327–368. Swan, T. (1956). “Economic growth and capital accumulation”. Economic Record 32, 334–361. Swank, D. (1996). “Culture, institutions, and economic growth”. American Journal of Political Science 40, 660–679. Swartz, S., Welsch, R. (1986). “Applications of bounded-influence and diagnostic methods in energy modeling”. In: Belsley, D., Kuh, E. (Eds.), Model Reliability. MIT Press, Cambridge. Tan, C.M. (2004). “No one true path to development: uncovering the interplay between geography, institutions, and ethnic fractionalization in economic development”. Mimeo, Tufts University. Tavares, J., Wacziarg, R. (2001). “How democracy affects growth”. European Economic Review 45 (8), 1341– 1378. Taylor, C. (1998). Socrates. Oxford University Press, New York. Temple, J. (1998). “Robustness tests of the augmented Solow model”. Journal of Applied Econometrics 13 (4), 361–375. Temple, J. (1999). “The new growth evidence”. Journal of Economic Literature 37 (1), 112–156. Temple, J. (2000a). “Inflation and growth: Stories short and tall”. Journal of Economic Surveys 14 (4), 395– 426. Temple, J. (2000b). “Growth regressions and what the textbooks don’t tell you”. Bulletin of Economic Research 52 (3), 181–205. Temple, J. (2003). “The long-run implications of growth theories”. Journal of Economic Surveys 17 (3), 497–510. Temple, J., Johnson, P. (1998). “Social capability and economic growth”. Quarterly Journal of Economics 113 (3), 965–990. Toya, H., Skidmore, M., Robertson, R. (2003). “Why are estimates of human capital’s contribution to growth so small”. Mimeo, Nagoya City University.
Ch. 8: Growth Econometrics
677
Wacziarg, R. (2002). “Review of Easterly’s The elusive quest for growth”. Journal of Economic Literature 40 (3), 907–918. Wacziarg, R., Welch, K. (2003). “Trade liberalization and growth: New evidence”. National Bureau of Economic Research Working Paper No. 10152. Warner, A. (1992). “Did the debt crisis cause the investment crisis?”. Quarterly Journal of Economics 107 (4), 1161–1186. Welsch, H. (2003). “Corruption, growth and the environment: A cross country analysis”. Mimeo, German Institute for Economic Research. White, H. (1980). “A heteroskedastic-consistent covariance matrix estimator and a direct test for heteroskedasticity”. Econometrica 48, 817–838. Whiteley, P. (2000). “Economic growth and social capital”. Political Studies 48, 443–466. Zak, P., Knack, S. (2001). “Trust and growth”. Economic Journal 111, 295–321. Zietz, J. (2001). “Heteroskedasticity and neglected parameter heterogeneity”. Oxford Bulletin of Economics and Statistics 63 (2), 263–273.
Chapter 9
ACCOUNTING FOR CROSS-COUNTRY INCOME DIFFERENCES FRANCESCO CASELLI* LSE, CEPR, and NBER e-mail:
[email protected]
Contents Abstract 1. Introduction 2. The measure of our ignorance 2.1. Basic data 2.2. Basic measures of success 2.3. Alternative measures used in the literature 2.4. Sub-samples
3. Robustness: basic stuff 3.1. Depreciation rate 3.2. Initial capital stock 3.3. Education-wage profile 3.4. Years of education 1 3.5. Years of education 2 3.6. Hours worked 3.7. Capital share
4. Quality of human capital 4.1. Quality of schooling: Inputs 4.1.1. Teachers’ human capital 4.1.2. Pupil–teacher ratios 4.1.3. Spending 4.2. Quality of schooling: test scores 4.3. Experience 4.4. Health 4.5. Social vs. private returns to schooling and health
5. Quality of physical capital 5.1. Composition 5.2. Vintage effects 5.3. Further problems with K
* A data set is posted at http://econ.lse.ac.uk/staff/caselli_francesco/.
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01009-9
680 681 683 685 686 688 689 690 690 691 693 694 694 695 696 698 698 699 701 702 703 706 708 710 711 711 715 716
680
F. Caselli
6. Sectorial differences in TFP 6.1. Industry studies 6.2. The role of agriculture 6.3. Sectorial composition and development accounting
7. Non-neutral differences in technology 7.1. Basic concepts and qualitative results 7.2. Development accounting with non-neutral differences
8. Conclusions Acknowledgements References
717 718 719 724 727 727 734 737 738 738
Abstract Why are some countries so much richer than others? Development Accounting is a first-pass attempt at organizing the answer around two proximate determinants: factors of production and efficiency. It answers the question “how much of the cross-country income variance can be attributed to differences in (physical and human) capital, and how much to differences in the efficiency with which capital is used?” Hence, it does for the cross-section what growth accounting does in the time series. The current consensus is that efficiency is at least as important as capital in explaining income differences. I survey the data and the basic methods that lead to this consensus, and explore several extensions. I argue that some of these extensions may lead to a reconsideration of the evidence.
Ch. 9: Accounting for Cross-Country Income Differences
681
1. Introduction This chapter is about development accounting. It is widely known, and will be found again to be true here, that cross-country differences in income per worker are enormous. Development accounting uses cross-country data on output and inputs, at one point in time, to assess the relative contribution of differences in factor quantities, and differences in the efficiency with which those factors are used, to these vast differences in per-worker incomes. Hence, it is the same idea of growth accounting (illustrated by Jorgenson’s chapter in this Handbook), with cross-country differences replacing cross-time differences. Conceptually, development accounting can be thought of as quantifying the relationship Income = F (Factors, Efficiency).
(1)
Like growth accounting, this is a potentially useful tool. If one found that Factors are able to account for most of the differences, then development economics could focus on explaining low rates of factor accumulation. There would of course be ample scope for controversy over the policies better suited to engineering higher investment rates in various types of capital, but there would be consensus over the fact that the intermediate goal of development policy is to engineer those higher rates. Instead, should one find that Efficiency differences play a large role, then one would have to confront the additional task of explaining why some countries extract more output than others from their factors of production. Experience suggests that this additional question is the hardest to crack. The consensus view in development accounting is that Efficiency plays a very large role. A sentence commonly used to summarize the existing literature sounds something like “differences in efficiency account for at least 50% of differences in per capita income”. The next section of this chapter (Section 2) will survey the existing literature, replicate its basic findings, and update them with more recent data. Looking at a crosssection of 94 countries in the year 1996, I confirm that standard procedures assign to Efficiency the role of the chief culprit. Operationally, the key steps in development accounting are: (1) choosing a functional form for F , and (2) accurately measuring Income and Factors. Efficiency is backed out as a residual. As for the Solow residual, this residual is a “measure of our ignorance” on the causes of poverty and under-development. And, as in growth accounting, one potentially promising research strategy is to try to “chip away” at this residual by improving on steps (1) and (2), i.e. by looking at alternative functional forms, and by attempting a more sophisticated measurement of Income and Factors. For example, one could try to include information on quality differences in the capital stock – instead of relying exclusively on quantity.
682
F. Caselli
The bulk of this chapter aims at outlining strategies for such a chipping away.1 It investigates the potential for different functional forms, and different ways of estimating inputs and outputs, to reduce the measure of our ignorance. Rather than reaching firm conclusions, it tries to classify ideas into more or less promising. Its contribution is to formulate sentences such as “improvements in the measurement of x are unlikely to significantly reduce the unexplained component of per-capita income differences”, or “the unexplained component is somewhat sensitive to the measurement of z, so this is a potentially fruitful area for further research”. The experiments I perform fall in five broad categories. The first is a fairly mechanical set of robustness checks with respect to the choice of parameters in the basic model used in the literature, as well as with respect to possible measurement errors in output, labor, and years of schooling. I conclude that none of these robustness checks seriously calls into question the conclusions from the current consensus (Section 3). Second, I consider extensions of the basic development-accounting framework aimed at improving the measurement of human capital. In most development-accounting exercises differences in human capital stem exclusively from differences in the quantity of schooling. One set of extensions I consider exploits cross-country data on school resources and test scores as proxies for the quality of education, and then uses these quality indicators to augment the quantity-based measure of human capital. I find that taking into account schooling quality leads to trivially small reductions in the measure of our ignorance. Another extension replicates existing work that augments human capital by a proxy of the health status of the labor force. There is some indication that this may lead to a significant reduction in the unexplained component of income, but I argue that the bulk of the variance most likely remains unexplained. All the measures of human capital considered are built on the assumption that the private return to human capital accurately describes its social return. I conclude this section with a brief discussion of why and how one may want to try and relax this assumption (Section 4). Third, I turn to the measurement of physical capital. Here I review contributions that highlight enormous cross-country variation in the composition of the stock of equipment. A simple model shows how to relate variation in capital composition to unobserved quality differences in the capital stock. How much of the responsibility for efficiency differences can be assigned to these differences in the quality of capital depends on parameters that are very hard to pin down, but the potential is extremely large. I therefore conclude that the composition of capital should be a key area of future research. I also glance at vintage-capital models, but argue that they hold little promise for development accounting, as well as at the distinction between private and public investment, which is instead potentially quite important (Section 5). The most innovative contributions of the chapter are represented by the fourth and fifth sets of extensions. In the former I explore the role of the sectorial composition of 1 The analogy in spirit with Jorgenson’s monumental contribution in growth accounting – some of which is collected in Jorgenson (1995a, 1995b) is obvious, but it stops there: the reader should expect nothing like the same level of depth, comprehensiveness, and insight.
Ch. 9: Accounting for Cross-Country Income Differences
683
output. The large differences in overall efficiency that are found at the aggregate level could reflect large differences in efficiency within each sector of the economy, but they could also be due to the fact that some countries have more of their inputs in intrinsically less productive sectors than others. I explore this idea by looking at an agriculture/nonagriculture decomposition (poor countries have as much as 90% of their workforce in agriculture, rich countries as little as 3%), but find that only a small fraction of the overall variation in efficiency is due to differences in sectorial composition: Efficiency differences appear to be a within industry phenomenon (Section 6). The last set of exercises explores a radical departure from the standard framework, and finds radically different answers. In the standard framework, which relies on a Cobb–Douglas specification of the production function, efficiency differences are factor neutral: if a country uses physical capital efficiently, it also necessarily uses human capital efficiently. I argue that this is a pretty restrictive assumption, and propose a simple CES generalization of the basic framework where cross-country efficiency differences are allowed to be non neutral. Stunningly, I find that, when non neutrality is allowed for, the data say that poor countries use physical capital more efficiently than rich countries (while rich countries use human capital more efficiently). Furthermore, when the development-accounting exercise is performed in a context of non-neutral efficiency differences the conclusions on the contribution of these differences to cross-country income inequality become very fragile. In particular, if the elasticity of substitution between physical and human capital is low enough, observed differences in factor endowments become able to explain the bulk of the cross-country income variance. I therefore conclude that the most important outstanding question in development accounting may well be what this elasticity of substitution is (Section 7). Before plunging into the data and the calculations, it is worthwhile to stress the limits of development accounting. Development accounting does not uncover the ultimate reasons why some countries are much richer than others: only the proximate ones. Like growth accounting, it has nothing to say on the causes of low factor accumulation, or low levels of efficiency. Indeed, the most likely scenario is that the same ultimate causes explain both. Furthermore, it has nothing to say on the way factor accumulation and efficiency influence each other, as they most probably do. Instead, it should be understood as a diagnostic tool, just as medical tests can tell one whether or not he is suffering from a certain ailment, but cannot reveal the causes of it. This does not make the test any the less useful.
2. The measure of our ignorance The key empirical result that motivates this chapter is that in a simple framework with two factors of production, physical and human capital, a large fraction of the crosscountry income variance remains unexplained. This result has been established by a variety of authors using a variety of techniques. Knight, Loayza and Villanueva (1993), Islam (1995), and Caselli, Esquivel and Lefort (1996), for example, used panel-data
684
F. Caselli
techniques to estimate (1). They all found that, after controlling for factor accumulation, country-specific effects played a large role in output differences, and interpreted these fixed effects as picking up differences in efficiency. King and Levine (1994), Klenow and Rodriguez-Clare (1997), Prescott (1998), and Hall and Jones (1999), instead, used a calibration approach, and found that plausible parametrizations of (1) had limited explanatory power without large efficiency differences. These studies used cross-country national-account data on inputs and outputs, but Hendricks (2002) was able to reach similar conclusions by using earnings of migrants to the United States, and Aiyar and Dalgaard (2002) by using a dual approach involving factor prices rather than quantities. All these papers were inspired by – and written in response to – the challenge posed by the seminal contribution of Mankiw, Romer and Weil (1992).2 In this section I revisit the basic development-accounting finding. Because I want to set the stage for a variety of extensions of the basic model, I adopt the calibration approach, which offers more flexibility in experimenting with different parameter values and functional forms.3 I adopt as the benchmark Hall and Jones’ production function, according to which a country’s GDP, Y , is Y = AK α (Lh)1−α ,
(2)
where K is the aggregate capital stock and Lh is the “quality adjusted” workforce, namely the number of workers L multiplied by their average human capital h. α is a constant. Clearly this is a special case of (1), where the residual A represents the efficiency with which factors are used. It is also clear that A corresponds to the standard notion of Total Factor Productivity (TFP), so until further notice I will speak of efficiency and TFP interchangeably. In per-worker terms the production function can be rewritten as y = Ak α h1−α ,
(3)
where k is the capital labor ratio (k = K/L). We want to know how much of the variation in y can be explained with variation in the observables, h and k, and how much is “residual” variation, i.e. must be attributed to differences in A. Clearly to answer this question we need, besides data on y, data on k and h, as well as a value for the capital share α.
2 However, there are some pre-1990s antecedents. In particular, the nine-country studies of Denison (1967), and Christensen, Cummings and Jorgenson (1981). 3 An earlier survey of the material covered in this section is provided by McGrattan and Schmitz (1999). See also Easterly and Levine (2001) for a review of development accounting as well as other evidence for cross-country efficiency differentials.
Ch. 9: Accounting for Cross-Country Income Differences
685
2.1. Basic data The basic data set used in this chapter combines variables from two sources. The first is version 6.1 of the Penn World Tables [PWT61 – Heston, Summers and Aten (2002)], i.e. the latest incarnation of the celebrated Summers and Heston (1991) data set. From PWT61 I extract output, capital, and the number of workers. The second is Barro and Lee (2001), which I use for educational attainment. Several additional data sources will be brought to bear for specific exercises in later sections, but the data we construct here will be crucial to everything we do. Previous authors have mostly used version 5.6 of the Penn World Tables (PWT56). They have therefore attempted to explain the world income distribution as of the late 1980s. By using version 6.1 I am able to update the basic result to the mid-90s. I measure y from PWT61 as real GDP per worker in international dollars (i.e. in PPP – this variable is called RGDPWOK in the original data set).4 I generate estimates of the capital stock, K, using the perpetual inventory equation Kt = It + (1 − δ)Kt−1 , where It is investment and δ is the depreciation rate. I measure It from PWT61 as real aggregate investment in PPP.5 Following standard practice, I compute the initial capital stock K0 as I0 /(g +δ), where I0 is the value of the investment series in the first year it is available, and g is the average geometric growth rate for the investment series between the first year with available data and 1970. The rationale for this choice is tenuous: I /(g + δ) is the expression for the capital stock in the steady state of the Solow model. We will see below whether results are very sensitive to this assumption, or for that matter to the others I am about to make, such as the one for δ, which – following the literature – I set to 0.06. To compute k, I divide K by the number of workers.6 To construct human capital I take from Barro and Lee (2001) the average years of schooling in the population over 25 year old. Following Hall and Jones (1999) this is turned into a measure of h through the formula: h = eφ(s) , where s is average years of schooling, and the function φ(s) is piecewise linear with
4 Some authors subtract from the PWT measure of GDP the value-added of the mining industry, because
not doing so would result in some oil-rich countries being among the most productive in the world. This rationale is inherently dubious (then why not subtracting the value added of agriculture and forestry, that also use natural resources abundantly?). More importantly, since a similar correction is not feasible for the capital stock, this procedure must result in hugely downward biased estimates of the TFP of these countries. I apply no such correction here. 5 Computed as RGDPL · POP · KI, where RGDPL is real income per capita obtained with the Laspeyres method, POP is the population, and KI is the investment share in total income. 6 Obtained as RGDPCH · POP/RGDPWOK, where RGDPCH is real GDP per capita computed with the chain method.
686
F. Caselli
slope 0.13 for s 4, 0.10 for 4 < s 8, and 0.07 for 8 < s.7 The rationale for this functional form is as follows. Given our production function, perfect competition in factor and good markets implies that the wage of a worker with s years of education is proportional to his human capital. Since the wage–schooling relationship is widely thought to be log-linear, this calls for a log-linear relation between h and s as well, or something like h = exp(φs s), with φs a constant. However, international data on education–wage profiles [Psacharopoulos (1994)] suggests that in Sub-Saharan Africa (which has the lowest levels of education) the return to one extra year of education is about 13.4 percent, the World average is 10.1 percent, and the OECD average is 6.8 percent. Hall and Jones’s measure tries to reconcile the log-linearity at the country level with the concavity across countries. s is observed in the data every five years, most recently in 2000. Since s moves slowly over time, a quinquennial observation can plausibly be employed for nearby dates as well. I treat a country as having “complete data” at date t if it has an uninterrupted investment series between 1960 and t, and it has an observation for s in 1995.8 With this definition, there are 94 countries with complete data in 1995, 94 in 1996, 91 in 1997, 90 in 1998, 87 in 1999, and 82 in 2000 (and 0 thereafter). Hence, I focus on 1996 as the most recent year that affords the largest sample. In this sample, for more than half of the countries the investment series starts in 1950.9 As is well known, per-capita income differences are enormous. The richest country in the sample (USA) has income per worker equal to 57,259 1996 international dollars, while the poorest (Zaire, today’s Democratic Republic of the Congo) has 630 – a ratio of 91. The ratio between the 90th (Canada) and the 10th percentile (Togo) of the income distribution, a measure of dispersion I’ll use prominently in the rest of the paper, is 21. The log-variance, another measure I’ll rely on heavily, is 1.30. For the last ingredient required by Equation (3), α, I (implicitly) use US time-series data on the capital-share, whose long-run (and roughly constant) average value is 1/3. All these data choices will be subject to scrutiny in the rest of the chapter – indeed, this scrutiny is one of the chapter’s contributions. 2.2. Basic measures of success With data on k, h, and y, and a choice for α, Equation (3) is one equation in the unknown A. In particular, after defining yKH = k α h1−α , we can rewrite (3) as y = A yKH ,
(4)
where both y and yKH are observable. I will refer to yKH as the factor-only model. 7 Specifically we have φ(s) = 0.134 · s if s 4, φ(s) = 0.134 · 4 + 0.101 · (s − 4) if 4 < s 8, φ(s) = 0.134 · 4 + 0.101 · 4 + 0.068 · (s − 8) if 8 < s. 8 Availability of data on income and labor force are not binding given these constraints. 9 Nicaragua in 1979 has I < 0, which we deal with by re-setting it to I = 0. Haiti has missing data on I in 1966, which we deal with by imputing the average of 1965 and 1967.
Ch. 9: Accounting for Cross-Country Income Differences
687
Throughout this chapter I will pursue the following version of the developmentaccounting question: how successful is the factor-only model at explaining crosscountry income differences? In other words, I will compare the (observed) variation in yHK to the (observed) variation in y. Clearly, this means that I am asking the following question. Suppose that all countries had the same level of efficiency A: what would the world income distribution look like in that case, compared to the actual one? To perform this assessment, I will look at two alternative measures. The first one is in the tradition of variance decompositions. From (4) we have var log(y) = var log(yKH ) + var log(A) + 2cov log(A), log(yKH ) . (5) Now notice that if all countries had the same level of TFP we would have var[log(A)] = cov[log(A), log(yKH )] = 0. Hence, a first measure of success of the factor-only model is var[log(yKH )] . success1 = var[log(y)] In our data the counterfactual variance, var[log(yKH )], takes the value 0.5. Since the observed variance of log(y), var[log(y)], is 1.30 this approach leads to the conclusion that the fraction of the variance of income explained by observed endowments is success1 = 0.39. While success1 is nicely grounded in the tradition of variance decomposition, it has the well-known drawback that variances are sensitive to outliers. A measure that is less sensitive to outliers is a measure of the inter-percentile differential. Define x p the value of the pth percentile of the distribution of x. My second measure of success of the factor-only model is 90 /y 10 yKH KH , y 90 /y 10 i.e. it compares what the 90th-to-10th percentile ratio would be in the counterfactual 90 /y 10 world with common technology, to the actual value. In the data the value of yKH KH 90 10 is 7. Since y /y is 21, according to the percentile ratio the fraction of the crosscountry income dispersion explained by observables is success2 = 0.34. I summarize the baseline experiment in Table 1. Clearly by both measures of success the dispersion of yKH is much less than the dispersion of y, and this is the basic fact that motivates this study.10 Before proceeding, it is useful to check that these results are consistent with the slightly different data used in previous studies. Using the Hall and Jones data set success1 is 0.40 (vs. 0.39 with ours), and success2 is 0.34, as with ours. As is evident, the different decade, country coverage, and methodology in assembling the PWT does not lead to important changes in this basic finding.
success2 =
10 Of course variation in y KH – even though much less than variation in y – is economically significant and
interesting in its own right. For recent studies shedding light on the sources of variation in k and h see, e.g., Bils and Klenow (2000), Hsieh and Klenow (2003), and Gourinchas and Jeanne (2003).
688
F. Caselli Table 1 Baseline success of the factor-only model var[log(y)] var[log(yKH )] success1
1.297 0.500 0.39
y 90 /y 10 90 /y 10 yKH KH success2
21 7 0.34
2.3. Alternative measures used in the literature success1 essentially asks what would the dispersion of (log) per-capita income be if all countries had the same level of efficiency, A, and then compares this counter-factual dispersion to the observed one. Klenow and Rodriguez-Clare (1997) propose the alternative measure: successKR =
var[log(yKH )] + cov[log(A), log(yKH )] , var[log(y)]
which differs from success1 for the covariance term in the numerator. In terms of Equation (5) successKR is equivalent to a variance decomposition in which the contribution from the covariance term is split evenly between A and yKH . Because in the data cov[log(A), log(yKH )] is positive (0.28) the Klenow and Rodriguez-Clare measure assigns a greater role to k and h than the simple ratio of variances: successKR is 0.60. Here I do not emphasize this measure because it does not answer the question: what would the dispersion of incomes be if all countries had the same A? As Klenow and Rodriguez-Clare explain, it asks the different question: “when we see 1% higher y, how much higher is our conditional expectation of yKH ?” which in my opinion is not as intuitive. Klenow and Rodriguez-Clare (1997) and Hall and Jones (1999) also work with a different version of the expression for per-capita income, because they rewrite (3) as a 1 k 1−α y= hA 1−α , y i.e. in terms of the capital–output ratio instead of the capital–labor ratio. In other words a their counterfactual income estimates based on factor-differences is y˜KH = ( yk ) 1−α h (instead of yKH = k α h1−α ). I find yKH more intuitive and cleaner, as y˜KH is not invariant to differences in A (since A affects y), and is therefore less appropriate for answering the question: “what would the income distribution look like if all countries had the α same A?”. Indeed, it is easy to see that y˜KH = yKH A α−1 . Whether var[log(y˜KH )] is greater or less than var[log(yKH )] depends on the relative magnitudes of (appropriately weighted) var[log(A)] and cov[log(yKH ), log(A)], with log(y˜KH ) getting less credit the (relatively) larger is the covariance. Intuitively, when A and yKH covary a lot, if the latter is very small the former is also very small, so that y˜KH does not vary as much. In practice this is indeed what happens: when using y˜KH the factors only model looks
Ch. 9: Accounting for Cross-Country Income Differences
689
even more unsuccessful than when using yKH : success1 is as low as 0.22, and success2 is 0.20. Notice that relative to Klenow and Rodriguez-Clare (1997) we have made two methodological changes whose effects go in opposite directions: omitting the covariance term from success1 lowers the explanatory power of factors, while writing y in terms of the capital–labor ratio increases it. This is why we end up with results that are in the same ball park. It is worth noting that Hall and Jones’ production function, Equation (2), is substantially more restrictive than the one used by some of the other authors in the literature. In particular Mankiw, Romer and Weil (1992) and Klenow and Rodriguez-Clare (1997) work with Y = K α H β L1−α−β . Equation (2) is the special case where β = 1 − α. The great advantage of the Hall and Jones’ formulation is that it generates the log-linear relation between wages and years of schooling that we exploited to calibrate h.11 Since wage data do seem to call for log-linear wage-education profiles, Hall and Jones’ restriction may be justified. 2.4. Sub-samples It may be interesting to take a look at the values that the success measures take in subsample of countries. This is done in Table 2, where I report success1 – as well as its two component parts – for the sub-samples of countries below and above the median per worker income; in and out of the OECD; and for the various continents. I also for convenience repeat the full-sample values. I do not report success2 because the small sample sizes make this variable hard to interpret. Obviously the variation in log income per worker is smaller the smaller and more homogeneous the sub-samples. Perhaps more interestingly, it is also smaller in subTable 2 Success in sub-samples Obs.
var[log(y)]
var[log(yKH )]
success1
Above the median Below the median
47 47
0.172 0.624
0.107 0.254
0.620 0.407
OECD Non-OECD
24 70
0.083 1.047
0.050 0.373
0.606 0.356
Africa Americas Asia and Oceania Europe
27 25 25 17
0.937 0.383 0.673 0.136
0.286 0.179 0.292 0.032
0.305 0.468 0.434 0.233
All
94
1.297
0.500
0.385
Sub-sample
11 With the Mankiw, Romer and Weil formulation the wage of a worker with s years of schooling is w(s) =
wL + wH h(s), where wL is the wage paid to “raw” labor and wH is the wage per unit of human capital.
690
F. Caselli
samples that tend to be richer on average (Above the median, OECD, Europe and Americas). It is indeed remarkable that, within the four continental groupings, the greatest variation in living standards is observed in Africa, a continent that is often depicted as flattened out by unmitigated and universal blight. The success of the factor-only model is higher in the above the median and in the OECD samples than in the below the median and non-OECD samples, respectively. Hence, it is easier to explain income differences among the rich than among the poor. Furthermore, as indicated by comparison with the results for the full sample, it is easier to explain income differences among the rich than between the rich and the poor – while it is roughly as easy to explain within-poor differences as rich-poor differences. At the continental level, success is highest in the Americas, with roughly 50% of the log income variance explained, and lowest in Europe, with 23%. The latter result is entirely driven by the inclusion of the lone eastern European country (Romania), whose very high level of human capital makes it difficult to explain its very low income [Caselli and Tenreyro (2004) generalize this finding to a broader sample]. When Romania is excluded the success of the factor-only model for Europe is virtually perfect. In sum, the factor-only model works the worst where we need it most: i.e. when poor countries are involved.12
3. Robustness: basic stuff The rest of this chapter is essentially about the robustness of the findings reported in Table 1. In this section I start out with a set of relatively straightforward and somewhat plodding robustness checks. In particular, I look at some of the parameters of the basic model as well as at some issues of measurement error. Subsequent subsections deviate from the benchmark increasingly aggressively. 3.1. Depreciation rate The effect of varying the depreciation rate in the perpetual-inventory calculation is to change the relative weight of old and new investment. A higher rate of depreciation will increase the relative capital stock of countries that have experienced high investment rates towards the end of the sample period. Poorer countries have in general experienced a larger increase in investment rates over the sample period, but the relative gain is very small, so it is unlikely that higher or lower depreciation rates will have a considerable impact on our calculation.13 In Figure 1 I compute and plot success1 and success2 for different values of δ. Clearly, the sensitivity of the factor-only model to changes in δ is minimal. 12 Results for Asia are virtually unchanged when I do not aggregate it with Oceania. 13 I computed the average investment rate in the sub-periods 1969–1972 and 1993–1996. Then I subtracted
these two averages, and correlated the resulting change in investment with real GDP per worker in 1996. The result is a modest −0.08.
Ch. 9: Accounting for Cross-Country Income Differences
691
Figure 1. Depreciation rate and success.
3.2. Initial capital stock The capital stocks in our calculations depend on the time series of investment (observable) and on assumptions on the initial capital stock, K0 , which is unobservable. Does the initial condition for the capital stock matter? One way to approach this question is to compute the statistic (1 − δ)t K0 , + ti=0 (1 − δ)i It−i
(1 − δ)t K0
(6)
i.e. the surviving portion of the guessed initial capital stock as a fraction of the final estimate of the capital stock. For t = 1996 the average across countries of this statistic is 0.01, with a maximum of 0.09 (Congo). This is prima facie evidence that the initial guess has very small “persistence”. However, this statistic is considerably negatively correlated with per capita income in 1996 (correlation coefficient −0.24), indicating that our estimate of the capital stock is more sensitive to the initial guess in the poorer countries in the sample. This may be troublesome because if we systematically overestimated the initial (and hence the final) capital stocks in poor countries, we will bias downward the measured success of the factor-only model. Furthermore, it is not implausible that our guess of the initial capital stock will be too high for poor countries. While rich countries may have roughly satisfied the steady state condition that motivates the assumption K0 = I0 /(g + δ), most of the poorer countries almost certainly did not.
692
F. Caselli
Indeed it is quite plausible that their investment rates were systematically lower before than after date 0 (i.e., before investment data became available for these countries).14 A first check on this problem is to focus on a narrower sample with longer investment series. If we focus only on the 50 countries with complete investment data starting in 1950, we should be fairly confident that the initial guess plays little role in the value of the final capital stock. In this smaller, and probably more reliable, sample we get success1 = 0.39, and success2 = 0.48. Hence, the ratio of log-variances is unchanged relative to the full 94-country sample, but the inter-percentile ratio shows a considerable improvement. Clearly, though, as the sample size declines the inter-percentile ratio becomes less compelling as a measure of dispersion, so on balance these results – though inconclusive – are reasonably reassuring.15 Another strategy is to attempt to set an upper bound on the measures of success, by making extreme assumptions on the degree to which the capital stock in poor countries is mismeasured. One such calculation assumes persistent growth rates in investment, I (as opposed to persistent investment levels). For example, we can construct a counterfactual investment series from 1940 to 1950 by assuming that the growth rate of investment in this period was the same as in the period 1950–1960. For countries with investment data starting after 1950 we can use the growth rate of investment in the first ten years of available data, and project back all the way to 1940. We can then use the perpetual inventory model on these data [always with K0 = I0 /(g + δ)], and measure success. On the full sample this yields success1 = 0.39 and success2 = 0.34, and on the sub-sample with complete I data starting in 1950 it yields success1 = 0.39 and success2 = 0.48, i.e. no change.16 Another experiment is to estimate the initial capital stock by assuming that the factoronly model adequately explained the data at time 0. Suppose that we trusted the estimate K0 = I0 /(g + δ) for the United States (where date 0 is 1950), and consequently for all other dates. Then for any other country we could estimate K0 by solving the expression Y0 K0 α L0 h0 1−α = , YUS KUS LUS hUS where 0 is now the first year for which both this country’s and the US’ data on investment, GDP, and human capital is available. Note that everything is observable in this equation except K0 (tough this does require us to construct new estimates of the human-capital stock for years prior to 1996). Clearly this procedure implies enormous variance in K0 , and this variance should persist to 1996, giving the factor-only model a 14 Circumstantial evidence that this may be the case is that a regression of the growth rate of total investment
between 1950 and 1960 on per-capita income in 1950 yields a statistically significant negative coefficient. 15 In this 50-country sample the variance of log income is 1.06, and the inter-percentile ratio is 8.5, i.e.
according to both measures there is less dispersion than in the full sample (1.3 and 21), but much more so for the inter-percentile differential. 16 When this method is used only to fill-in data between 1950 and 1961 it yields success = 0.38 and 1 success2 = 0.34.
Ch. 9: Accounting for Cross-Country Income Differences
693
real good shot at explaining the data. On the full sample this yields success1 = 0.41 and success2 = 0.34, and on the sub-sample with complete I data starting in 1950 it yields success1 = 0.45 and success2 = 0.48. Hence, even when the initial capital stock is constructed in such a way that the factor-only model fully explained the data at time 0, the model falls far short in 1996. I conclude from this set of exercises that improving the initial capital stock estimates is not likely to lead to major revisions to the baseline result. 3.3. Education-wage profile By assuming decreasing aggregate returns to years of schooling the Hall and Jones method dampens the variation across countries in human capital, thereby potentially increasing the role of differences in technology. More generally, our measure of human capital may obviously be quite sensitive to the parameters of the function φ(s). One way of checking this is to assume a constant rate of return, or φ(s) = φs s, and experiment with various values of the (constant) return to schooling φs . Since countries with higher per-capita income have higher average years of education, the factor-only model will be the more successful the steeper is the education–wage profile. Figure 2 confirms this by plotting success1 and success2 as functions of φs . While higher assumed returns to an extra year of education do lead to greater explanatory power for the factor-only model, only returns that are implausibly large lead to substantial successes. For success1 (success2 ) to be 0.75 the return to one year of school-
Figure 2. Returns to schooling and success.
694
F. Caselli
ing would have to be around 25% (26%). As already mentioned, in the Psacharopoulos (1994) survey the average return is about 10%. The highest estimated return is 28.8% (for Jamaica in 1989), but this is a clear outlier since the second-highest is 20.1% (Ivory Coast, 1986). These tend to be OLS estimates. Instrumental variable estimates on US data are 17–20 percent at the very highest [Card (1999)] – sufficient for our measures of success to just clear the 50 percent threshold. But the IV estimates tend to be lower in developing countries. For example, Duflo (2001) finds instrumental-variable estimates of the return to schooling in Indonesia in a range between 6.8% to 10.6% – and roughly similar to the OLS estimates. It seems, then, that independently of the return to schooling, the variation in schooling years across countries is too limited to explain very large a fraction of the cross-country variation in incomes.17 3.4. Years of education 1 De la Fuente and Doménech (2002) survey data and methodological issues that arise in the construction of international educational attainment data, such as the average years of education in the Barro and Lee data set. Their conclusion – perhaps not surprisingly – is that such series are rather noisy, and that this explains in part why human-capital based models often perform rather poorly. For several OECD countries they also construct new estimates that take into account more comprehensive information than is usually exploited, and find that for this restricted sample their measure substantially improves the empirical explanatory power of human capital. To see if incorrect measurement of s is a likely culprit for the lack of success of the factor-only model, I compute our success statistics for the sub-sample covered in the De la Fuente and Doménech (2002) data set, first with our baseline data, and then with the new figures provided by these authors (in their Appendix 1, Table A.4) for 1995. This data is available for only 15 of our 94 countries. In this 15-country sample, with our baseline (Barro and Lee) schooling data, I obtain success1 = 0.487 and succeess2 = 0.976. In the same sample with the De la Fuente and Doménech data success1 = 0.490 and success2 = 0.977. The differences seem small. This result is not particularly surprising because De la Fuente and Doménech (2002) show that the discrepancies between their measures and the ones in the literature are (1) stronger in first differences than in levels; and (2) stronger at the beginning of the sample than at the end. Indeed, for the 15-country sample in 1995 the correlation between the De la Fuente and Doménech (2002), and the Barro and Lee (2001) data I use in the rest of the paper is 0.78. Incorrect measurement of s is not the reason why the factor-only model performs poorly. 3.5. Years of education 2 So far we have used Barro and Lee’s data on years of schooling in the population over 25 years old. This may be appropriate for rich countries with a large share of college 17 This discussion, of course, assumes away human-capital externalities. I return to this in Section 4.5.
Ch. 9: Accounting for Cross-Country Income Differences
695
graduates. But it is much less appropriate for the typical country in our sample. Barro and Lee (2001) also report data on years of schooling for the population over 15 years of age. These data can be combined with the data on the over-25 as follows. First, note that we can write 1 s15 = (N15 − N25 )s<25 + N25 s25 , N15 where s15 and s25 are the average years of education in the population over 15 and over 25 years of age, respectively (the data), and s<25 is the (unknown) average years of education in the population between 15 and 25 years of age. N15 and N25 are the sizes of the population over 15 and over 25. With data on N15 and N25 , then, this is one equation in one unknown that can easily be solved for s<25 . With an estimate of s<25 at hand, one could then produce a new measure of s as s = L<25 s<25 + L25 s25 , where L<25 and L25 are estimates of the proportions of above and below 25 year olds in the economically active populations. I take data for N15 , N25 , L<25 , and L25 from LABORSTA, the data base of the International Labor Organization (ILO).18 This alternative measure of s can be constructed for 91 countries. With it, our success measures are 0.362 and 0.322, respectively. Hence, this potentially improved measure of human capital worsens its explanatory power. The reason is not hard to see: poor countries experienced much faster growth in schooling than rich countries. This means, in particular, that the education gap is much smaller for the cohort less than 25 years of age. Hence, bringing the education of this cohort into the picture reduces the cross-country variation in human capital. 3.6. Hours worked So far I have measured L as the economically active population, a measure that basically coincides with the labor force. Of course, the number of hours worked – the concept of labor input we would ideally like to use in our calculations – may be far from proportional to this measure, both because of cross-country differences in unemployment, and because the average employed worker may supply different amount of hours in different countries, for example because of different wages or different opportunity cost of work (in terms of forgone leisure). LABORSTA includes data on weekly hours for 41 countries in 1996, and in this sample the variance of hours worked is indeed very large: from 57 (Egypt), to 26.6 (Moldova).19 However, the particular cross-country pattern of these hours does not go 18 For both the population at large and the economically active subset the data is available at 10-year intervals
from 1950 to 2000, with the lucky exception that there is also an observation for 1995, which I use. The population is broken down in 5-year age intervals, so it’s a no brainer to aggregate up to the numbers we need. 19 For the subset of 28 countries that are both in the ILO sample and in our baseline sample the maximum is the same, and the minimum is 31.7 (Netherlands).
696
F. Caselli
Figure 3. Hours worked around the world.
in the direction that favors the factor-only model. Figure 3, where I plot weekly hours against log income per worker (for countries with data on both variables), clearly shows that workers work fewer hours in high-income countries.20 This implies that – if anything – TFP differences are under-estimated! In principle, this effect may be compensated – and possibly reversed – by higher unemployment rates in poorer countries. Figure 4 plots data from the World Bank’s World Development Indicators (WDI) on unemployment rates against log per capita income, in 1996.21 Contrary to common perceptions, unemployment rates are not higher in poorer countries. It therefore seems unlikely that further pursuing differences in hours worked may lead to a significant improvement in the explanatory power of the factoronly model. 3.7. Capital share The exponents on k and h act as weights: the larger the exponent on, say, k, the larger the impact that variation in k will have on the observed variation in y. However, under constant returns to scale these exponents sum to one, so increasing the explanatory power of k through increases in α also means lowering the explanatory power of h. 20 The coefficient of a regression of log weekly hours on log per-worker income implies that a one percent
increase in per worker income lowers weekly hours by about 0.11% – which is sizable. 21 ILO data on unemployment generate a very similar picture.
Ch. 9: Accounting for Cross-Country Income Differences
697
Figure 4. Unemployment rates around the world.
Figure 5. Capital share and success.
Because k is more variable across countries than h, in general one can increase the explanatory power of the “factor-only” model by increasing α. Figure 5 plots success1 and success2 as functions of the capital share α. As predicted, the fit of the factor-only model increases with the assumed value of α. Remarkably,
698
F. Caselli
our measures of success are quite sensitive to variations in α. For example, a relatively minor increase of α to 40% is sufficient to bring success1 to 0.5, and a 50% capital share implies success measures in the 0.6–0.7 range. Success is almost complete with a 60% capital share. This high sensitivity of the success measure, especially around the benchmark value of α = 1/3, imply that the parameter α is a “sensitive choice” in development accounting, and that our assessment of the quantitative extent of our ignorance may change non-trivially with more precise measures of the capital share. Still, as long as the capital share is below 40%, most of the variation in income is still explained by TFP.
4. Quality of human capital We have seen that simple parametric deviations from the benchmark measurements in Section 2 do not alter the basic conclusion that differences in the efficiency with which factors are used are extremely large. Here and in the next section I subject this claim to further scrutiny, by investigating possible differences in the quality of the human and physical capital stocks. For, the measures adopted thus far are exclusively based on the quantity of education and the quantity of investment, but do not allow, for example, one year of education in country A to generate more human capital than in country B. Similarly, they do not allow one dollar of investment in country A to purchase capital of higher quality than in country B. I conceptualize differences in the quality of human capital by writing h = Ah eφ(s) . Up until now, I have assumed that Ah is constant across countries. In this section I examine the possibility that Ah is variable. 4.1. Quality of schooling: Inputs Klenow and Rodriguez-Clare (1997) and Bils and Klenow (2000) have proposed ways to allow the quality of education to differ across countries. Their main focus is that the human capital of one generation (the “students”) may depend on the human capital of the preceding one (the “teachers”). One can further extend their framework to allow for differences in teacher–pupil ratios, and other resources invested in education. For example, one could write: φ
φ
Ah = p φp mφm kh k ht h ,
(7)
where p is the teacher–pupil ratio, m is the amount of teaching materials per student (textbooks, etc.), kh is the amount of structures per student (classrooms, gyms, labs, . . . ), and ht is the human capital of teachers: the better the teachers, the more students will get out of their years of schooling. More generally, the term ht might capture externalities in the process of acquiring human capital.
Ch. 9: Accounting for Cross-Country Income Differences
699
In this sub-section I will try to plug in values for the inputs p, m, kh , and ht , and calibrate the corresponding elasticities. Unfortunately, little is known about the latter. Indeed, they are the object of intense controversy in and out of academe. Hence, I will typically look at a fairly broad range of values. 4.1.1. Teachers’ human capital I begin by focusing on the last of the factors in (7), ht . To isolate this particular channel for differences in schooling quality I ignore other sources, i.e. I set φp = φm = φk = 0, which is essentially Bils and Klenow’s assumption. When we review the evidence on these other φs, we’ll see that this assumption may actually be quite realistic. If we make the additional “steady state” assumption that ht = h, we can write h = eφ(s)/(1−φh ) , and plugging this into (3) we get: y = Ak α e
(1−α)φ(s) 1−φh
.
(8)
Note that this formulation magnifies the impact of differences in years of schooling, the more so the larger the elasticity of student human capital to teacher’s human capital. I continue to choose α = 1/3, and the function φ(·) as described in Section 2. The new, unknown parameter is φh . In Figure 6 I plot success1 and success2 as functions of
Figure 6. φh and success.
700
F. Caselli
this parameter. Note that φh = 0 is the baseline case of Section 2. At the low values of φh implied by the baseline case the success measures are fairly insensitive to changes in the elasticity of students’ to teachers’ human capital. However, the relationship between the success measures and φh is sufficiently convex that when φh is 69% success is complete. Coincidentally, 69% is very close to the upper bound of the range of values Bils and Klenow consider “admissible” for φh (67%), though clearly this admissibility is purely theoretical: their preferred values are actually in the 0–20% range.22 One way to think about what is reasonable for φh is to compute by how much the teachers’ human capital effect “blows up” the Mincerian return: from Equation (8) we see that with φh = 0.2 the “social” return to schooling is 1.2 times the private one; with φh = 0.4 it is 1.7 times larger; and with φh = 0.67 it is 3 times more. While it is hard to reach a firm conclusion, it would seem that reasonable priors on φh are inconsistent with large improvements in the fit of the factor-only model. Turning to possible objective estimates of φh , the first option is of course to look for estimates of the effect of teachers’ years of education on student achievement. This is because under our assumptions differences in teacher’s quality are ultimately determined by teachers’ years of education. However, Hanushek’s (2004) review of the literature concludes that teachers’ measurable credentials – including years of education – have no measurable impact on schooling outcomes.23 Another way to formulate priors on the possible magnitude of φh is to look at evidence on the effect of parental education on wages. After all, our simple representativeagent model of human capital is not explicit about the particular way the economy’s average level of human capital enhances the learning experience of new members of society. We can legitimately re-interpret ht , therefore, as the human capital of parents. One recent set of log-wage regressions including the schooling of parents (alongside with an individual’s own schooling) is presented in Altonji and Dunn (1996). Depending on data sources, and on whether the regression is estimated for men or women, their coefficient on father’s years of schooling ranges from −0.5% to 1%, and the coefficient on mother’s schooling from less than 0.1% to about 0.5%. Note that given our
22 They compute this upper-bound (roughly) as follows. Given data on schooling years of different cohorts,
given a Mincerian wage-years of schooling profile, and given a value for φh , it is possible to estimate the growth rate of h, and hence the contribution of growth in h to the growth of y. Holding the Mincerian profile constant, the larger φh , the larger the fraction of growth explained by human capital (for reasons already touched upon in the text). For Bils and Klenow the upper bound for φh is the value such that growth in human capital explains all of growth – or the value beyond which the residual, growth in TFP, would have to be negative. When the Mincerian profile features decreasing returns, as in our baseline specification, and as in Bils and Klenow’s preferred specification, this maximum value for φh is 0.19; when the Mincerian profile is linear the maximum becomes 0.67. The decreasing returns case allows for a smaller maximum φh because, towards the beginning of the sample period, many countries with very low education levels have very high Mincerian returns, implying fast growth in human capital. 23 This does not mean that teachers’ quality does not matter, of course. It only means that teacher quality is not related to measurable credentials. This unmeasurable quality effect remains (appropriately) a part of the measure of our ignorance.
Ch. 9: Accounting for Cross-Country Income Differences
701
functional form assumption the coefficient of parental education is φs φh , where φs is the return to own years of schooling (assumed constant for simplicity). If the return to own schooling, φs , is in the ball park of 0.10 (as the evidence on Mincerian coefficients roughly implies), and we focus on Altonji and Dunn’s upper bound of 0.01 for φs φh , we conclude that φh cannot be more than 0.1. A quick check with Figure 6 reveals that even this upper bound does not support a meaningful boost in the explanatory power of overall human capital.24 4.1.2. Pupil–teacher ratios φ
The term ht h in Equation (7) does not appear to enhance the success of the factoronly model. I now consider the term p φp . Lee and Barro (2001) report data on the pupil–teacher ratio in a cross-section of countries for various periods since 1960, and separately for primary and secondary schooling. For each country, I focus on the pupil– teacher ratio in the years when the average worker attended school. To pinpoint this year, I need to start with an estimate of the age of the average worker, which I construct from LABORSTA.25 Then I assume that children begin primary schooling at the age of 6. This implies that the relevant observation for the primary pupil–teacher ratio would be for the year 1996-age + 6. Furthermore, using unpublished panel data by Barro and Lee on the duration of primary and secondary schooling, we can determine the relevant observation for the secondary pupil–teacher ratio as 1996-age + 6 + duration of primary school. In order to combine the primary and secondary ratios in a unique statistic, I combine the duration of schooling data with our basic data on the average years of schooling of the population over 25 years of age, s, to determine what fraction of schooling time the average worker spent in primary, and what fraction in secondary school. I then construct p by simply averaging the primary and secondary teacher–pupil ratio using as weights the time spent in these two grades, respectively. At the end of all this, I have data on p for 87 of our 94 countries.26 24 Another way to boost the contribution of human capital to income would be to assume that parental/teacher
human capital increases the slope and not just the intercept of the log-wage – schooling relation. This is indeed Altonji and Dunn’s main focus. However, they do not find much evidence in support of this hypothesis. 25 As already mentioned, LABORSTA breaks down the economically active population in 5-year age intervals, from 10–14, to 60–64, plus a catch-all bracket for 65+. To get at the average age of a worker I simply weighted the middle year of each interval by the fraction of the labor force in that interval. For the 65+ group, I arbitrarily used 68. Of my 94-country sample, this data is available for 91 countries. I imputed average age for the two missing countries (Taiwan and Zaire) through a cross-sectional regression of average age of worker on per-worker income and years of schooling. 26 Since the pupil–teacher ratio is observed at five-year intervals in practice we “target” the observation closest to the estimated age at which the average worker went to school. With this procedure, in the sample of 86 countries with data on pupil–teacher ratios, the target dates for primary school attendance are 1960 for two countries, 1965 for 40, and 1970 for 44. For secondary school attendance the target dates are 1965 (one country), 1970 (25), 1975 (55), and 1980 (5).
702
F. Caselli
Figure 7. φp and success.
Figure 7 plots success1 and success2 as functions of φp . Since richer countries have higher teacher–pupil ratios, clearly a higher elasticity of human capital to this ratio implies a better fit, or greater success. What is a reasonable range of values for φp ? At the low end of the spectrum there is the position taken by Hanushek and coauthors, who conclude that resources – including a large teacher–pupil ratio – have little if any effect on economic outcomes.27 At the other end of the spectrum, my own reading of the literature indicates that the highest published estimate of φp is a very sizable 0.5.28 However, even with this extremely high estimate it is clear that the fit of the model improves modestly, with our success measures barely attaining even the 50% mark. 4.1.3. Spending I do not have direct data on materials, m and structures per student, kh . Instead, I have – always from Lee and Barro (2001) – a measure of government spending per student in PPP dollars. The bulk of this spending typically goes to teacher salaries, so variation in these data also reflect differences in the number and possibly the quality of teachers per 27 In a cross-country context, Hanushek and Kimko (2000) find no evidence that more resources improve
schooling quality, and Hanushek, Rivkin and Taylor (1996) and Hanushek (2003) reach the same conclusion upon reviewing the US-based literature. 28 Card and Krueger (1996). I infer this number from their reported 5% increase in earnings associated with a 10% reduction in class size for white men.
Ch. 9: Accounting for Cross-Country Income Differences
703
Figure 8. φsp and success.
student. However, to a certain extent, they may also reflect variation in materials. For the purposes of using these data, it seems sensible, therefore, to replace Equation (7) by Ah = spendingφsp , where the dating of the spending observation and the weights given to primary and secondary spending are determined as for the pupil–teacher ratio. For this exercise, I have data for 64 countries, and for this sample the measures of success are plotted in Figure 8. Again, rich countries devote more resources to education per student, so the fit of the model improves with φsp . However, again, there is the Hanushek position in the papers cited above, according to which φsp should be thought of as close to zero. At the other end of the range I have found an estimate of 0.2, which clearly is barely sufficient to even clear the 50% threshold of explanatory power.29,30 4.2. Quality of schooling: test scores Another way to investigate the potential of quality-of-education modifications to the basic model is to exploit information on the performance of students on reading, science, 29 Johnson and Stafford (1973), who run a regression of log hourly wages on log state expenditure per student
(and controls), obtaining a coefficient of 0.198. For the reasons discussed by Hanushek and co-authors there is a high presumption of upward bias in this estimate. 30 Lee and Barro (2001) also report information on the duration of the schooling year (in days and hours), but these variables – while highly variable – are weakly, and if anything negatively, correlated with per-capita income, so that they are highly unpromising from the perspective of improving the fit of the model. Similarly, teacher salaries, as a percent of per-capita GDP, are higher in poorer countries.
704
F. Caselli
and math tests in different countries. When students in one country outperform students of another (holding grade constant), we can assume that they have enjoyed schooling of higher quality, whether this higher quality comes from higher teacher–pupil ratios, quality of teachers, other expenditures, or other unobservables specific to the production of human capital. Hanushek and Kimko (2000) find that test scores enter significantly in growth regressions. To implement this idea I think of Ah as a function of test scores: higher test scores signal higher human capital. Suppose, for example, that the relationship between school quality and test results is given by Ah = eφτ τ , where τ is the test score.31 Then, with data on test scores, if we knew φτ we could construct a new counterfactual measure of yKH , or the output attributable to “observable” factors of production. I use data on test scores provided by Lee and Barro (2001), who for several countries observe data on multiple tests (e.g.: math, science, and reading), and for multiple grades, at different dates. Ideally I would follow the procedure outlined in the previous subsection, i.e. to “target” the year in which the average worker is presumed to have been in school. Because this data is very sparse, however, and mostly available in recent dates, I will focus on recent observations. This procedure is appropriate if the quality of education has grown over time at roughly similar rates across countries. The two tests that afford the greatest country coverage – 28 countries with overlapping test, input, and output data – are a math and a science test imparted to 13-year-old children between 1993 and 1998. The scores are standardized on a 0-to-100 scale, and I take the simple average of the two test scores.32 With this summary measure of τ at hand, in Figure 9 I plot our measures of success against φτ . The result should be treated with great caution given the very small sample size. Notice for example that, even for φτ = 0, both measures of success are considerably higher than in the full sample. With that caveat, it is true that students in rich countries perform better in standardized tests, and therefore the success of the model improves with φτ . To find a benchmark for φτ against which to evaluate Figure 9, notice that our assumption on the relationship between test scores and school quality translates into an assumption on the relation between test scores and wages: a unit increase in test scores is associated with a φτ proportional increase in wages. I have chosen this exponential form because studies of the relationship between test scores and wages tend to report coefficients from regressions of log wages on absolute test scores. For example, the coefficient φτ ×100 (after rescaling the test data to be in the same units as ours) is reported to be between 0.08 and 0.34 by Murnane, Willett and Levy (1995); between 0.12 and 0.27 by Currie and Thomas (1999), and between 0.55 and 1.02 by Neal and Johnson (1996) – which is at the high end of the range of available estimates.33 31 The reason for the exponential form will be apparent below. 32 The correlation between the two test scores is 0.87. 33 Murnane, Willett and Levy (1995) report the coefficient of the regression of log of weekly wages on math
test scores (tested in senior high school) to vary between 0.00004 and 0.00017 (depending on sex and cohorts
Ch. 9: Accounting for Cross-Country Income Differences
705
Figure 9. φτ (×100) and success.
Inspection of Figure 9 given this range of values suggests that using test scores as proxies for schooling quality cannot substantially improve the performance of the factor-only model. The problem is that, given the drastically reduced sample size, it is hard to take a stand on the degree to which this finding generalizes. I can attain a slight increase in sample size if I drop the requirement that the tests be imparted in roughly the same period and roughly the same subject. If I use all the test considered, US data). Since the test results are reported to vary between 2 and 17 points, we assume that the test is on a 0–20 scale. When translated to our 0–100 scale this implies the φs reported in the text. The Currie and Thomas (1999) results imply that “students who score in the upper quartile of the reading exam earn 20% more than students who score in the lower quartile of the exam, while students in the top quartile of the math exam earn another 19% more. When they control for father’s occupation, father’s education, children, birth order, mother’s age, and birth weight, the wage gap between the top and bottom quartile on the reading exam is 13% for men and 18% for women, and on the math exam it is 17% for men and 9% for women” [Krueger (2003, p. 25)]. From here we can infer that φτ varies between 0.0012 and 0.0027 (dividing the percentage change in the wage by the 75 points that separate the top from the bottom quartile). Neal and Johnson (1996) run a regression of log real yearly wages on standardized AFQT test scores, and find a coefficient between 0.17 and 0.29. Introducing more controls the coefficients are between 0.12 and 0.16. Since the standard deviation of AFQT scores (as reported in the note to their Appendix A.3) is 36.65, this implies that a one-point increase in AFQT scores increases wages by between 0.33 and 0.79 percent. Given that AFQT scores range between 95 and 258, this implies a φ between 0.0055 and 0.0102 (treating each of the AFQT points as 1.64 of our 100 points). (Whether AFQT scores are measures of schooling outcome is somewhat controversial.) Hanushek and Kimko (2000) use essentially the same international test scores we are using here to explain the earnings of migrants to the US, and obtain φτ × 100 of approximately 0.2.
706
F. Caselli
Figure 10. φτ (×100) and success, all tests in the 90s.
scores available from the 1990s, i.e. I average across all tests irrespective of subject, age group, and specific year, our sample size becomes 42 and success is given by Figure 10. If I use all available tests, including those from decades before the 1990s, the sample size is 45 and success is shown by Figure 11. As we increase the sample size, the potential success of the factor-only model if anything declines. 4.3. Experience Klenow and Rodriguez-Clare (1997) and Bils and Klenow (2000) also allow for differences across countries in experience levels. Since Mincerian wage regressions indicate that experience increases earnings, it makes sense to correct human capital for the contribution of experience. This correction has two conflicting effects on the explanatory power of human capital. Since workers in rich countries live longer than workers in poor countries, this should boost rich countries’ human capital. However, since richcountry workers spend more time in school, a smaller proportion of their time is spent accumulating experience, which reduces their relative human capital.34 Klenow and Rodriguez-Clare (1997) find that the net effect is negative: experience is actually higher in poor countries. Hence, in their calculations correcting for experience lowers the explanatory power of the factor-only model. However, in order to compute
34 A third effect, that adds to the second, is that rich-country workers may retire earlier.
Ch. 9: Accounting for Cross-Country Income Differences
707
Figure 11. φτ (×100) and success, all available test scores.
the average age of workers they rely on UN data on the age structure of the population, while in principle it would be more accurate to look at the age structure of the labor force. Using again the LABORSTA-based measure of the average age of the economically active population in the formula experience = age-schooling − 6, I find that the correlation between experience and per-capita income is −0.29 in our 94-country sample. Therefore, I confirm the Klenow and Rodriguez-Clare conclusion that poor countries have less education but more experience. Adding experience to the factor-only model, therefore, will only worsen its explanatory power.35
35 This discussion assumes implicitly that experience enters linearly in the production function for human
capital, an assumption we know not to be valid. However, for this consideration to overturn the conclusion we just reached, it would have to be the case that poor countries are to the right of the argmax, which seems very unlikely: in my data, the maximum average experience is 27 years. More importantly the discussion also abstracts from compositional issues. Feyrer (2002) uncovers an economically important and remarkably robust association between a country’s productivity and its share of the labor force that is between 40 and 49 years of age. Extending the development accounting framework to capture this effect would be a worthwhile task.
708
F. Caselli
4.4. Health Weil (2001) and Shastry and Weil (2003) point out that there are very large crosscountry differences in nutrition and health status, and argue that these differences map into substantial differences in energy and capacity for effort. They find that accounting for health differences across countries increases by one-third the explanatory power of human capital for differences in per-capita income. Weil (2001) uses as a proxy for health the Adult Mortality Rate (AMR), which measures the fraction of current 15-year-old people who will die before age 60, under the assumption that age-specific death rates in the future will stay constant at current levels. In practice, this is a measure of the probability of dying “young”, and is therefore a plausible (inverse) proxy for overall health status. The correction of human capital for health can be implemented through the assumption Ah = eφamr AMR , where clearly φamr < 0: a higher adult mortality rate implies a less energetic workforce. I gather cross-country data on AMRs from the WDI, covering 92 of our 94 countries, for the year 1999. I plot success for different values of −φamr in Figure 12. Since richer countries have healthier workers, the explanatory power of human capital increases in −φamr . Weil’s preferred value for −φamr (×100) is 1.68. Conditional on this value, I do confirm his finding that the factor-only model’s explanatory power improves considerably – indeed by almost one third, taking us well above the 50 percent threshold of success. This is therefore a very important and promising contribution.
Figure 12. −φamr and success.
Ch. 9: Accounting for Cross-Country Income Differences
709
Given his choices of functional form, however, this calibration implies that a onepercentage-point reduction in the probability of dying young is associated with a 1.68 percent increase in human capital, and hence in wages. Put another way, reducing the probability of dying before the age of 60 (as of age 15) by 6 percentage points has the same impact on wages as one extra year of schooling. This effect may seem a bit too large to be realistic. Given the somewhat tortuous – if ingenious – path through which Weil comes up with this calibration, I would tend to consider this number an upper bound.36 One can perhaps improve on Weil’s exercise by exploiting as a proxy for health the information on average birthweight generated by Behrman and Rosenzweig (2004). The great advantage is that these authors also report estimates (based on within twin-pair regressions) of the economic returns to higher birthweight (as measured by wages). This should provide a more solid base for calibration. There are, however, several shortcoming. First, birthweight may be less strongly correlated with health than the adult mortality rate. Second, the cross-section of mean birthweights refers to new borns in 1989, so it captures (a correlate of) the health of a cohort of workers that was not even in the labor force (aside from the most extreme cases of child labor) as of the date of our development accounting exercise (1996). Hence, one needs to assume a very high degree of persistence in cross-country differences in birthweights in order to put a lot of stock in this exercise. Third, the point estimate of the returns to birthweight are from a sample of US female twins, and one may question their applicability to the population at large.37 With those caveats, Figure 13 plots the usual measures of success when we assume Ah = eφbw BW , where BW is Behrman and Rosenzweig’s mean birthweight (in pounds), and φbw is the elasticity of human capital to birthweight. The number of countries is 83. Since birthweight is higher in rich countries a higher value of φbw increases the explanatory power of human capital. However, the value of φbw implied by Behrman and Rosenzweig’s log wage regressions is 0.076, which implies a trivially small improvement in the success of the factor-only model. This is substantively the same conclusion of Behrman and Rosenzweig, who report that the variance of φbw BW is less than one percent of the variance of log(y). In sum, while the results with the adult mortality rate strongly imply that a correction for differences in health status is a first-order requirement in the measurement of human capital, those using birthweight are much less supportive. In light of the shortcomings of
36 Weil uses published micro-level estimates from three developing countries to infer the elasticity of human
capital to height. He then uses time series data from Korea and Sweden to estimate a relationship between height and the AMR. He then combines these two pieces of information to infer the elasticity of human capital to the AMR. In essence, he is using the AMR to predict height, and then applies to the predicted height the microeconomic estimate of the effect of height on wages. 37 Fourth, Behrman and Rosenzweig also find some evidence of non-linearity in the birthweight-log wage relationship, so to perform a really accurate exercise one should use data on each country’s entire birthweight distribution. Of course that is easily available for only a handful of countries.
710
F. Caselli
Figure 13. φbw and success.
both exercises, however, it seems highly worthwhile to try and explore the matter further with more accurate indicators of health and more precisely calibrated parameters. 4.5. Social vs. private returns to schooling and health Some additional important caveats about the nature of the calculations above is in order before I “set aside” human capital. Recall that the function φ(s) that we have used to map years of schooling into human capital was calibrated on estimates of private rates of return. Similarly, attempts at calibrating the health-human capital relation rely on observed private returns to health. But, as pointed out by various authors, and especially forcefully by Pritchett (2003), these private returns may bear little relationship to the social (or aggregate) return to education, which is of course what one would like to plug in our calculations. As Pritchett points out, the social return to education may be higher or lower than the private one. Most growth theorists instinctively think about the former case, as they have in mind models with positive spillovers from human capital. However, Pritchett’s review of the evidence is typical in finding very little empirical support for positive externalities.38 On the other hand, various versions of the education-as-signalling-device model, as well as models of rent seeking, imply that the social return to education is
38 See, e.g., Heckman and Klenow (1997) and Acemoglu and Angrist (2000).
Ch. 9: Accounting for Cross-Country Income Differences
711
lower than the private return.39 This possibility is quite compelling. Note, however, that our calculations above imply that if we uniformly lower the social rate of return to education, cross-country schooling inequality will explain even less of income inequality than it does in our benchmark calculation (see Figure 2). Pritchett, however, also convincingly argues that the extent of rent seeking, and therefore the extent to which the social return is below the private return, is much larger in poor countries. For example, in many poor countries the government employs an overwhelmingly large share of college graduates. This is sometimes the result of guaranteedemployment rules that commit the government to find employment to anyone with a tertiary degree. In contrast, in rich countries most college graduates work in the private sector. Since standard rent-seeking arguments imply that the government sector is intrinsically likely to make less efficient use of resources, this implies that on average the social return to education will be lower in poor countries. This effect is of course reinforced by the fact that poor countries are notoriously more prone to corruption and rent seeking than rich ones. This will help. If the social rate of return to education (and health) is allowed to be higher in rich countries, then the variance of h will increase, and with it the explanatory power of the model. How important this could be quantitatively is hard to say, but by all means it would be worth finding out. A first exploratory step may be to break down the labor force into government-employees and private-sector workers. One may then retain the parameterization of the benchmark case for the private sector workers, but assume lower returns for government employees.
5. Quality of physical capital 5.1. Composition Recent research by Eaton and Kortum (2001) has shown that most of the world’s capital is produced in a small number of R&D-intensive countries, while the rest of the world generally imports its equipment. This suggests that, for most countries, (widely available data on) imports of capital of a certain type are an adequate proxy for overall investment in that type of equipment. Caselli and Wilson (2004) exploit this observation to investigate cross-country differences in the composition of the capital stock.40 Their results – a partial summary of which is shown in Table 3 – are startling: different types of equipment constitute widely varying fractions of the overall capital stock across countries. For each of the nine equipment categories Caselli and Wilson work with, the share in total investment in 1995 has minima in the low single digits, and maxima
39 See, e.g., Murphy, Shleifer and Vishny (1991), or Gelb, Knight and Sabot (1988). 40 This idea is also used in Caselli and Coleman (2001a). The equipment-import data are extracted from the
Feenstra (2000) data set.
712
F. Caselli Table 3 The composition of the capital stock Fabricated NonOffice Electrical Comm. Motor- Other Aircraft Prof. metal electrical computing equipment equipment vehicles transp. goods products machinery accounting
Mean STD Min Max Corr Y R&D
0.08 0.06 0.01 0.55 −0.25 202
0.21 0.08 0.03 0.48 −0.14
0.06 0.05 0.01 0.41 0.53
0.14 0.07 0.01 0.59 0.27
0.11 0.05 0.01 0.37 0.20
0.24 0.10 0.01 0.55 −0.32
887
1170
848
2280
1810
0.03 0.04 0.00 0.34 −0.41
0.05 0.09 0.00 0.88 0.14
0.07 0.03 0.01 0.23 0.33
57
1880
801
that vary between 20 percent and 80 percent! The standard deviations of investment shares are always large relative to the cross-country means. Furthermore, this enormous heterogeneity is systematically related to per capita income, as the correlations with income of the various investment shares are large in absolute values. To begin to see why this vast heterogeneity in the composition of equipment may matter for development-accounting, Table 3 also reports global, cumulated R&D expenditures in the various equipment categories.41 The wide variation in R&D spending across types reinforces the impression of equipment heterogeneity across countries: since equipment shares vary so much, so does the embodied-technology content of the aggregate capital stock. If the R&D content of equipment determines its quality, i.e. its productivity per dollar of market value, one begins to suspect that the quality of capital – and not only its quantity – may vary across countries. Furthermore, these differences are systematic, since richer countries appear to employ high R&D capital much more than poor countries. A simple way to see this is to look at Figure 14, from Wilson (2004). For equipment types with a high R&D content the share in overall investment is positively correlated with output per worker, while the opposite is true for low-tech equipment types. Could it be that rich countries use higher quality equipment, and that this higher quality accounts for some of the residual TFP variance? To see how this may work it is useful to write down a very simple model. Imagine that final output Y is produced combining various intermediate inputs, xp , according to the CES production function42 P 1 γ γ Y =B (xp ) , γ < 1, p=1
41 This is sometimes referred to as the “R&D stock” of a certain type of capital, as it is computed by cu-
mulating past R&D spending with the perpetual inventory method. The R&D spending data come from the ANBERD data base. See Caselli and Wilson (2004) for details. 42 Production functions such as this one have been the staple of recent developments in growth theory.
Ch. 9: Accounting for Cross-Country Income Differences
713
Figure 14. Capital composition and income.
where B is a disembodied total factor productivity term. Intermediate-good p is produced combining equipment and labor: xp = Ap (hp Lp )1−α (Kp )α ,
0 < α < 1,
(9)
where Kp measures the quantity of equipment (in current dollars) used to produce intermediate-input p, hp Lp is human-capital augmented labor in sector p, and Ap is the productivity of sector p. The key assumption is that capital is heterogeneous: there are P distinct types of capital, and each type is product specific, in the sense that intermediate p can only be produced with capital of type p. In other words, an intermediate is identified by the type of equipment that is used in its production.43 The assumption that γ < 1 implies that – in producing aggregate output – all these activities are imperfect substitutes. The productivity term Ap is product specific. Product variation in A allows for the possibility that one dollar spent on equipment of type p may deliver different amounts of efficiency units if instead spent on type p . For example, the embodied-technology
43 For example, for equipment-type “trucks”, the corresponding intermediate good x (say, “road transporta-
tion”) is the one obtained by combining workers with trucks. For equipment-type “computers”, the corresponding intermediate good is “computing services”, etc. Hence, our intermediates do not easily map into industries or sectors (computers are used in most industries), but rather into the various types of activities (transport, computing, etc.) required to generate output within each sector.
714
F. Caselli
content of good p may be greater because the industry producing equipment of type p is more R&D intensive.44 A number of simplifying assumptions allow one to write a simple formula that brings the idea of the “quality of capital” in sharp relief. In keeping with the representativeagent spirit of the rest of the chapter say, then, that human capital per worker is constant across sector, i.e. hp = h, and that labor is free to flow across sectors so that the marginal product of labor is the same for all equipment types. Then one can show that the output equation can be rewritten as45 Y = (K)α (hL)1−α B
P
1−(1−α)γ (Ap )
γ 1−(1−α)γ
(ξp )
αγ 1−(1−α)γ
γ
,
(10)
p=1
where of course K is the total market value of the capital stock, L is total labor, and ξp is the share of capital of type p in the total capital stock, or ξp = Kp /K. Up until now we have been writing Y = (K)α (hL)1−α A, and we have struggled with the fact that A seems to play an enormously important role in determining output differences. The last equation neatly shows the relationship between A and the composition of the capital stock: if different types of capital have different productivities, then the observed wild variation in equipment shares ξ implies that the quality of capital – over and above its quantity K – can vary across countries and can account for a portion of the unexplained variation. Caselli and Wilson (2004) propose a regression-based approach to make inferences on the various Ap s. Unfortunately, even with knowledge of the Ap s it is virtually impossible to bound the amount of income variance that the ξp s can explain. This is because the last term of Equation (10) is exceedingly sensitive to the value of γ , and there seem at the moment to exist no reliable approach to the calibration of this elasticity. If γ is sufficiently low, i.e. capital types are sufficiently poor substitutes for one another, the quality of capital accounts for all of the unexplained component of income differences. However, whether such values are reasonable or not our current state of knowledge cannot say. I conclude therefore that further research on the composition of capital is an important priority for development accounting.
44 Since, for simplicity, we have written an aggregate production function that is symmetric in the services
provided by different types of capital, product variation in Ap may also reflect differences in the various capital types’ shares in aggregate output. Caselli and Wilson also allow the productivity term Ap to vary across countries, for a given equipment type p. The idea behind country variation is that equipment of type p may be more complementary with the characteristics of country i, an idea that is strongly supported by their empirical results. In order to make sure that cross-country differences in Kp measure physical differences in installed capital we need to assume that the law of one price holds. This is plausible, since we know that most capital is imported from a few world producers. If the law of one price does not hold, however, cross-country differences in Ap may also reflect price differences. 45 See Caselli and Wilson (2004) for details.
Ch. 9: Accounting for Cross-Country Income Differences
715
5.2. Vintage effects Solow’s (1959) paper on vintage capital formalized the idea that technological progress is embodied in capital goods. Rodriguez-Clare (1996), Jovanovic and Rob (1997), Parente (2000), Mateos-Planas (2000) have noted that this could potentially enhance the explanatory power of cross-country differences in investment rates. The idea, of course, is that low investment rates will be associated with lower adoption of new technology. Indeed, some of these authors have argued that versions of the capital-embodied model greatly outperform the homogeneous-capital model in accounting for cross-country income differences. Formally, the vintage-capital model could be described by the formulas: Yt = (Lt ht )1−α
t
α At−i Kt−i ,
i=0
(11)
Kt−i = (1 − δ)i I˜t−1 , where I˜t−1 is investment at time t −1 in terms of the consumption good. Consistent with Solow’s idea, this model has the property that capital installed by sacrificing one unit of consumption at date s yields As (1 − δ)t−s efficiency units of capital at time t, while capital installed by sacrificing one unit of consumption at time t yields At efficiency units. If As < At we have that the earlier sacrifice in consumption contributes less to output today not only because of physical depreciation, but also because of the older vintage. To appreciate the potential consequences of this note that the same comparison in a homogeneous-capital (i.e. disembodied technical change) model would be between At (1 − δ)t−s and At , i.e. differences in efficiency units obtained with the same sacrifice of consumption would only be due to physical depreciation. As explained by Greenwood, Hercowitz and Krusell (1997), under certain conditions the formulation above is equivalent to Yt = (Lt ht )1−α
t
α , K˜ t−i
i=0
K˜ t−i = (1 − δ)i qt−i I˜t−i , which has the following interpretation. Instead of one unit of consumption producing equal amounts of capital of increasing quality at different dates, one unit of consumption produces increasing amounts of capital of the same quality. Because the aggregate implications of growth in A are isomorphic to those of growth in q, the two formulations can be equivalent representations of the idea of embodied technological progress. The second version of the model suggests, however, that – at least in principle – the estimates of the physical capital stocks we have been using until now do actually already reflect embodied technical change. This is because the real investment series we construct from PWT61 is a series for real investment in terms of the investment good, and not in terms of the consumption good. In other words, the PWT61 investment data
716
F. Caselli
that we use are data on Is = qs I˜s , and not on I˜s . Therefore, vintage effects – or at least those vintage effects that show up in a reduced relative price of investment goods – should already be accounted for. As a very rough check on this argument, I have run a cross-country OLS regression of output per quality-adjusted worker on a distributed-lag function of depreciated investments (in units of the investment good). I.e. the left hand side variable was Yt /(Lt ht )1−α , and the right hand side variables were It , (1 − δ)It−1 , (1 − δ)2 It−2 , . . . . I experimented with 5, 10, and 20 lags of investment. The homogeneous-capital hypothesis – or, much more accurately, the hypothesis that all vintage effects are adequately captured by investment-good prices in PWT61 – is equivalent to all the coefficients taking the same value, irrespective of the vintage. The “vintage effects” hypothesis would predict that coefficients on recent lags of investment would be systematically larger than those on older lags. The result was somewhat inconclusive, in that both hypotheses were rejected: all coefficients were not statistically the same, but neither they fell monotonically with the lag of investment. A possible explanation is that the price deflators in PWT work well enough to remove systematic vintage effects, but the remaining i.i.d. measurement error occasionally makes some vintages look more productive than others. In any case there is little indication that vintage-based models will significantly improve on the benchmark. 5.3. Further problems with K The investment series I have used to estimate the capital stock is an aggregate of private and public investment expenditures. As Pritchett (2000) very convincingly argues, however, elementary logic and vast anecdotal experience suggest that many governments’ investment efforts are much less productive than private ones. There is an infinite supply of examples where government investments have not produced anything tangible (non-existent highways, industrial complexes that have never been completed, etc.). Furthermore, even when public investments do materialize, the resulting structures and machinery may be run less efficiently than under private management.46 As for Pritchett’s (2003) criticism of schooling-based measures of human capital based on private returns, his (2000) criticism of what he derogatorily but accurately calls CUDIE (cumulated depreciated investment expenditure) may help shedding light on the puzzles we are concerned with, both because governments tend to account for a larger share of production, employment, and capital ownership in poorer countries, and because less-accountable poor-country governments are likely to be disproportionately less efficient (relative to the private sector) than rich country ones. Hence, there are good reasons to expect the government to play an especially detrimental role in the
46 The 1994 World Bank’s World Development Report documents substantial cross-country differences in
the efficiency with which public infrastructure is used. Hulten (1996) uses these inefficiency indicators in a growth-regression exercise.
Ch. 9: Accounting for Cross-Country Income Differences
717
productivity of investment in poor countries. This implies that the “effective” variance of K is larger than in the baseline model.47 As I suggested in the previous section for the analogous problem with human capital, a first pass at investigating this issue would be to try to separate out public from private investment, and apply different weights in the perpetual-inventory calculation, which would become Kt = Iprivate,t + γ Ipublic,t − δKt−1 . One could then try to re-do development accounting with this modified capital measure (possibly for various values of γ ). Unfortunately, I have not been able to identify reliable and updated PPP breakdowns of the investment series into private and public capital.48 Perhaps a cleaner exercise, but also even more ambitious, would be to try to completely net out the government from the development accounting exercise. I.e. subtract the government’s share from aggregate output, capital ownership, and employment, and perform the development-accounting exercise on the residual (private) inputs and output. This confronts the same data limitations as the exercise described in the previous paragraph, and the additional problem of coming up with a reliable PPP government share of GDP.49
6. Sectorial differences in TFP Up to here in this chapter I have treated a country’s GDP as if it was produced in a single sector, i.e. as if GDP measured the physical output of a homogeneous good. The basic message has been that it is impossible to explain cross-country differences in income without admitting a large role for differences in TFP. It is tempting to jump to the conclusion that these TFP differences signal the existence of barriers to technology adoption in less developed countries, or other frictions that broadly make some countries “function” less well than others. But large differences in TFP could also be the result of variation in the weights in GDP of sectors with different sectorial-level productivity – even when these sectorial productivities are identical across countries. In this case
47 While the emphasis is on the role of the government, Pritchett gives various other reasons why CUDIE is
problematic. 48 However, both the WDI and the IMF’s Government Finance Statistics (GFS) have data on government
capital formation in domestic currency. One could therefore potentially construct a non-government investment series, and – in principle – use PWT deflators to turn this series into a real investment series. Whether one would get anything sensible out of this is another matter. 49 One can of course not simply subtract G from Y , in PWT, because G is government consumption of goods and services, and not government production. In the NIPA government output is defined as the sum of factor payments, i.e. compensation to general-government employees plus general-government consumption of fixed capital. Something like these categories are reported in the IMF’s Government Financial Statistics. Unfortunately, the spottiness of these data forced me to abandon this particular enterprise.
718
F. Caselli
we would want to focus on barriers to the mobility of factors across sectors, instead of barriers to the mobility of technology or work practices across countries.50 6.1. Industry studies There is a tradition of productivity comparisons at the industry, or even at the firm, level. Particularly illustrative of the advantages of this approach are a series of reports published by the McKinsey Global Institute. These studies focus on painstaking comparisons of the production functions (broadly construed) of narrowly defined industries (from automobile, to beer, to retail banking) in a few industrialized economies (mostly US, Japan, Germany, and the UK). Baily and Solow (2001) present a thoughtful survey of the achievements, as well as the shortcomings, of these studies (as well as extensive references).51 Briefly, even within narrowly defined industries, and even among countries at very similar levels of development, total factor productivity presents remarkable variation. A similar conclusion, for somewhat more aggregated manufacturing industries (but more countries), is reached by Harrigan (1997, 1999). Hence, industry-level studies suggest that aggregate TFP differences are not solely due to differences in the weights of high- and low-TFP sectors. Since these studies are often limited to industries in a few highly developed economies, however, one should be cautious before assuming that the same causes drive the low TFP levels of less developed economies. Besides confirming that TFP differences exist also at the industry level, the McKinsey researchers are often also able to shed some tentative light on their sources. In particular, they highlight differences in working practices, and they are sometimes also able to link inefficient practices to the regulatory environment. In general, the studies point to a link between the degree of competition domestic producers are exposed to (as affected by the amount of regulation), and the efficiency with which they organize their labor input. This is of course an important, and plausible, finding. Further support for this view comes from the work of Schmitz (2001) on the North-American iron-ore industry, which shows convincingly that the efficiency of labor practices is very responsive to the degree of product-market competition.52
50 Here we are arguing that differences in sectoral composition may account for some of the differences in the
level of GDP. Koren and Tenreyro (2004) show that these same differences account for a substantial fraction of differences in its volatility. 51 A precursor to these intra-industry cross-country productivity comparisons is the three-country study of Conrad and Jorgenson (1985). See also Wagner and van Ark (1996). 52 Another wonderful industry-level comparison of cross-country productivity differences is presented by Clark (1987), who examines the productivity of cotton mills around the world in the early years of the twentieth century. He shows that, assuming constant capital–labor ratios, the textile industries of Britain and New England would have had a huge cost disadvantage relative to India, Japan, and many other countries. Yet, British cotton textiles dominated export markets. Clark shows that the various countries’ industries used identical equipment, and that the expertise to organize and run the mills could not have differed too much. Rather,
Ch. 9: Accounting for Cross-Country Income Differences
719
6.2. The role of agriculture As mentioned, existing cross-country comparisons of sectorial TFP tend to be limited to small sets of developed countries. The goal of this section is therefore to provide a rough, preliminary assessment of the sectorial-composition interpretation of TFP differences that extends to developing countries as well. In particular, I will focus on an agriculture–nonagriculture split of GDP. The main reason for looking at this particular breakdown is easily inferred from Figure 15: in the poorest countries of the world virtually everyone works in agriculture, and in the richest virtually nobody does. It is obvious that this is the most important source of variation in the composition of GDP around the World. Another reason for focusing on agriculture is that I have no PPP output data for other sectors. Finally, the agriculture-nonagriculture dualism has traditionally played a central role in the history of thought on economic development.53 The main purpose of this section, then, is to assess the hypothesis that (i) agriculture is an intrinsically low TFP sector, and (ii) poor countries’ low aggregate TFP is due
Figure 15. The importance of agriculture.
the source of the productivity differences boils down to the fact that each English worker was willing to tend to a much larger number of machines. In low-productivity countries workers were idle most of the time. Why this was so remains a bit of a mystery, and one should be cautious in assuming that this finding would still hold up one century later. Nevertheless, Clark’s findings reinforce the case that labor practices may be an important source of observed differences in productivity. 53 Some of the classics are Fisher (1945), Clark (1940), Rostow (1960), Nurkse (1953), Lewis (1954), Kuznets (1966), and Jorgenson (1961).
720
F. Caselli
in substantial measure to their high shares of agriculture. In this subsection, however, I start by taking a preliminary (and perhaps somewhat digressive) look at basic data on agricultural GDPs, non-agricultural GDPs, and agricultural labor shares. What we find should provide further motivation for asking the development-accounting question with disaggregated data. I begin by writing per-worker GDP in PPP as y = PA yA lA + PA yA lA ,
(12)
where PA (PA ) is the international (PPP) price of agricultural (non-agricultural) goods, yA (yA ) is the per-worker output of the agricultural (non-agricultural) sector, and lA (lA ) is the agricultural (non-agricultural) share of employment. We already know that lA varies dramatically across countries. What about PA yA and PA yA ? The Food and Agriculture Organization (FAO) has collected and published crosscountry data on producer prices in agriculture for a large number of countries between 1970 and 1990 [Rao (1993), see also Restuccia, Yang and Zhu (2003)]. This permits the construction of PPP exchange rates for agriculture, and therefore of PPP comparisons of agricultural output. Going even further, the FAO researchers also assembled some data on agricultural inputs for the year 1985, and this allowed them to generate a cross-section of PPP agricultural GDPs. Furthermore, the methodology followed in the FAO study deliberately follows the methods of Summers and Heston (1991). Hence, the estimates of PPP agricultural GDP in the FAO data set are comparable to the aggregate PPP numbers of PWT61. The FAO data set also obviously contains information on agricultural employment (which was used for Figure 15). A difficulty that needs to be addressed, however, is that the FAO numbers for PPP agricultural GDP do not directly map into the quantity PA yA in Equation (12). The reason is as follows. The FAO PPP GDPs aggregate the quantities of the various agricultural products by a set of “international prices”, that are essentially weighted averages of each country’s prices. The same is done in the PWT for all goods and services. However, the two systems use a different normalization for the international prices – i.e. they have the same relative prices of agricultural products but different absolute levels. Hence, the PPP agricultural value-added coming from the FAO data set cannot simply be plugged into Equation (12) with PWT aggregate value added on the left hand side.54 54 Formulas may help here. Suppose that there is only one agricultural good and one nonagricultural good.
Call PA and PA their respective prices in PWT international dollars: the PWT’s unit of account. The normalization used in PWT is that PA yA,US + PA yA,US = PA,US yA,US + PA,US yA,US , where PA,US (PA,US ) is the price of A (A) in the US (this is what Summers and Heston mean when they say, somewhat opaquely, that the PPP of the US is 1). Instead, the normalization in the FAO data set is PAF yA,US = PA,US yA,US , where PAF is the price of the agricultural good in FAO’s international dollars. In this two-good example this obviously implies PAF = PA,US , so that PAF yA = PA yA only if we have PA = PA,US , as well. But the above-described normalization of PWT prices does not assure this at all, and indeed it would be true only by coincidence. I guess one could put this into PWT-speak, and say that the fact that the PPP of GDP is normal-
Ch. 9: Accounting for Cross-Country Income Differences
721
I try to solve this problem as follows. It is well known that – because they are quantity-weighted – international-dollar relative prices in the PWT closely resemble rich-country, and especially US, relative prices [Hill (2000)]. Hence, for the US, we should have PA,US YA,US PA YA,US ≈ , D YUS YUS D is the agricultural share in GDP at domestic prices, which can where PA,US YA,US /YUS D are known (the be obtained from the WDI. In these equations YUS , and PA,US YA,US /YUS former from PWT, and the latter from WDI). Hence, we can solve for PA YA,US . Now recall that the FAO estimates for PPP agricultural GDP differ from the (implicit) PWT estimates only by a constant of normalization. It should follow that if we rescale all of the FAO agricultural GDP numbers such that the US value coincides with PA YA,US (as just calculated) we have an estimate of the contribution of agriculture to PWT GDP that we can plug into Equation (12). As already pointed out by Restuccia, Yang and Zhu (2003), the most striking feature of the FAO data is that variation in agricultural value-added per worker (in PPP) dwarfs the variation in aggregate value-added per worker. In the largest sample with data on both agricultural and aggregate GDP per worker (80 countries) the inter-percentile range in agricultural GDP is 45 and the log-variance is 2.15. The corresponding numbers for aggregate GDP are 22 and 1.18, respectively.55 Real agricultural GDP per worker is plotted in Figure 16 against real aggregate GDP per worker. Subtracting real agricultural GDP from aggregate GDP, it is also possible to back out non-agricultural value-added per worker.56 The (not surprising but nonetheless) very important finding is that differences in labor productivity in the non-agricultural sector are much smaller than differences in aggregate labor productivity (and, a fortiori, in agricultural labor productivity). The inter-percentile range is only 4.16 (compared with 22 for aggregate GDP and 45 for agriculture) and the log-variance is 0.33 (compared with 1.18 and 2.15). Figure 17 plots non-agricultural value-added per worker against aggregate value-added per-worker. Comparison of Figures 16 and 17 shows that labor productivity is generally higher outside than inside agriculture, and this is much more true for developing countries, an observation previously made by Gollin, Parente and Rogerson (2000).57
ized to 1 for the US, does not imply that the PPP of individual sectors (such as agriculture) is normalized to 1 for the US as well. 55 Due to the high persistence of the World’s income distribution these last two numbers are very close to the corresponding numbers for our benchmark 1996 sample (21 and 1.3). 56 This is where the measurement problem described before becomes important: the log-variance of agricultural GDP is obviously insensitive to the price-normalization adopted, but non-agricultural GDP is computed as a residual from Equation (12), so it is crucial that PA yA and y are in the same units. 57 Only Australia and New Zealand have higher productivity in agriculture than in nonagriculture. The average log-difference between nonagricultural and agricultural output per worker in the entire 80-country sample is 2.18; among the poorest 20 it is 3.13; among the richest 20 it is 0.86.
722
F. Caselli
Figure 16. Labor productivity in agriculture.
Figure 17. Labor productivity outside of agriculture.
Ch. 9: Accounting for Cross-Country Income Differences
723
Table 4 Counterfactual World income distributions Variable Actual real output per worker Counterfactual 1: US yA , own yA & lA Counterfactual 2: US yA , own yA & lA Counterfactual 3: US lA , own yA & yA
log-variance
Int. range
1.18 0.04 0.58 0.34
22 1.6 7.0 4.2
Recalling now from Figure 15 that the third component of Equation (12), the employment share lA , ranges from almost 0 percent in the richest countries to almost 100 percent in the poorest, we conclude that poor countries have most of their labor force in the sector where they are particularly unproductive.58,59 We can summarize this first overview of the sectorial data by saying that there are three proximate reasons for poor countries’ poverty: their much lower labor productivity in agriculture; their somewhat lower labor productivity outside agriculture; and their larger share of employment in the sector that – on average – is less productive. To quantify these effects Table 4 presents income-dispersion statistics (log-variance and inter-percentile range) in the data (first row), and under alternative counterfactual assumptions on industry-level productivity and labor shares. Counterfactual 1 is that all countries have the US level of agricultural GDP per worker, but their own level of non-agricultural GDP per worker and agricultural labor share. Counterfactual 2 is that all countries have the US-level of non-agricultural GDP per worker, but their own level of agricultural GDP per worker and agricultural labor share. Counterfactual 3 is that all countries have the US agricultural labor share, but their own level of agricultural and non-agricultural GDP per worker.60
58 Attempts at explaining this apparent deviation from comparative advantage abound. It may be that the non-
agricultural sector has greater skill requirements, so that low human-capital economies are constrained in the supply of non-agricultural workers [Caselli and Coleman (2001b)]; or it could be that investment distortions push producers into the home (agricultural) sector [Gollin, Parente and Rogerson (2000)]; or it could be that economies are subject to a “subsistence constraint”, such that resources cannot start moving out of agriculture until agriculture is sufficiently productive to generate a surplus that will feed the industrial class [Gollin, Parente and Rogerson (2001); Restuccia, Yang and Zhu (2003)]; or it could be that some countries are “trapped” in agriculture by a coordination failure [long tradition; most recently Graham and Temple (2001)]. 59 Given the huge employment shares of agriculture in Figure 15 one would guess that in most developing countries agriculture would account for an equally vast share of GDP. In fact, the agricultural share of GDP is always below 40 percent. This is a consequence of the disproportionately low productivity of agriculture in low-income countries. Also note that the PPP agricultural share in GDP is both much less variable across countries, and – for most countries – lower than the domestic-currency agricultural share in GDP. This is because – perhaps contrary to common wisdom – the relative price of agricultural goods is higher in poor countries than in the US. 60 A similar calculation is reported by Restuccia, Yang and Zhu (2003). Calculations in this spirit can also be found in Caselli and Coleman (2001b) (for US regions), and Gollin, Parente and Rogerson (2001).
724
F. Caselli
The results are stunning. The figures in the second row imply that if poor countries achieved the same level of agricultural labor productivity as the US, world income inequality would virtually disappear! This is of course a reflection of the convergence of US agricultural incomes to US non-agricultural incomes [documented in Caselli and Coleman (2001b)], as well as the huge agricultural share of employment in many of the poorest countries. However, the other two counterfactual experiments also generate large declines in dispersion. Because agriculture is generally much less productive than non-agriculture, reducing the agricultural employment share to US levels would reduce income inequality by an enormous two thirds (third row). And cross-country non-agricultural productivity differences, while much less than agricultural ones, are still sufficiently large that income inequality would fall by about one half if poor countries were as productive outside of agriculture as the US (second row).61 6.3. Sectorial composition and development accounting The previous subsection establishes that there are very large within industry crosscountry differences in output per worker. Indeed, the agricultural GDP differences are substantially larger than the aggregate GDP ones. Furthermore, these cross-country differences in industry GDPs are seen to potentially “account” for a large fraction of the cross-country dispersion in aggregate income. We can now return to the original question: are these large differences in agricultural GDP attributable to the amounts of observable inputs employed in agriculture by the various countries, or are they the result of industry-level cross-country TFP differences? This is of course “the” developmentaccounting question. To try to answer this question, we need assumptions on the industry production functions, as well as ways of measuring industry-level inputs. I will assume that each of the two sectors produces according to a Cobb–Douglas technology. In agriculture, the factors of production are capital, labor, and land (T ). In non-agriculture, they are capital and labor: YA = AA (KA )αA (LA hA )βA (TA )1−αA −βA ,
(13)
YA = AA (KA )αA (LA hA )1−αA .
(14)
The goal now is to construct counter-factual agricultural and non-agricultural output data, β
1−αA −βA
αA A yA,HK = kA hA tA
yA,HK =
,
α 1−α k Ah A. A A
61 Of course, this discussion abstracts from the changes in world-wide agricultural relative prices that such
changes would bring about.
Ch. 9: Accounting for Cross-Country Income Differences
725
These counter-factual data answer the question: what would the world distribution of agricultural (non-agricultural) output per worker look like if all countries had the same agricultural (non-agricultural) total factor productivity? Assume that the rates of return on capital must be equalized across sectors – a plausible arbitrage condition. This is easily seen to imply αA
P D YA PAD YA = αA A , KA KA
where PAD (P D ) is the domestic producer price of agricultural (non-agricultural) goods, A which will generally differ from the PPP price, and is the price the domestic investor cares about (unless he produces for the export market). The quantities PAD YA and P D YA A are, of course, agricultural and non-agricultural output in domestic prices, and they are observable from WDI.62 Hence, combining this equation with KA + KA = K,
(15)
where K is the (observable) total capital stock, we can back out KA and KA , and hence (with labor shares calculated from FAO and PWT) kA and kA .63 It is harder to come up with numbers for hA and hA . Caselli and Coleman (2001b) show that there is a very systematic tendency for agriculture to be one of the least skill intensive sectors in the economy. For example, for each of the years between 1940 and 1990 they rank the roughly 120 industries featured in the US Census of Population by percentage of workers with an elementary degree or less, and in each of these years agriculture consistently ranks in the bottom 10. This suggests that it may not be unreasonable to set hA = 1 in all countries, i.e. that the agricultural work force is made up by workers with no education. It is then easy to compute human capital per worker outside of agriculture. In particular, if s is the average years of schooling in the labor force, and a fraction lA has no education and works in agriculture, then the remaining fraction 1 − lA must have years of education sA = s/(1 − lA ). hA is then computed with the “standard” formula linking years of education to human capital by way of Mincerian returns. For tA I simply plug in WDI data on each country’s endowment of arable land. For the parameters, I use essentially the same calibration as in Caselli and Coleman (2001b), which is in turn based on the work of Jorgenson and Gollop (1992). The labor share in the USA is about 60% in both farming and non-farming. The capital share in agriculture is about 21%, and the remainder is absorbed by land. 62 In fact, we use the share of agriculture in GDP in domestic prices from the WDI. This is sufficient for the calculation below since only the ratio YAD /Y D is needed. A 63 Instead of using a no-arbitrage condition for capital, sector specific capital inputs may be recovered using
an indifference condition for rural–urban labor flows. This condition would involve, among other things, an “urban wage premium” compensating for costs of skill acquisition [as, for example, in Caselli and Coleman (2001b)] or for the lower probability of finding a job, as in the celebrated model of Harris and Todaro (1970). See Temple (2003) for an interesting calibration of the Harris–Todaro model aiming to assess the output costs of the labor-market rigidities that lead to economic dualism.
726
F. Caselli Table 5 Success within sectors Sector Agriculture Nonagriculture
success1
success 2
0.15 0.59
0.09 0.63
For the 65 countries for which we can construct yA,HK and yA,HK and for which we have measures of yA and yA , the success of the factor-only model is as reported in Table 5. Once again, the results are striking. Briefly, the factor-only model explains virtually nothing of the observed per-capita income variance in agriculture: it’s entirely a story of TFP differences, even more so than for aggregate GDP. Conversely, physical and human capital inputs do a better than usual job at explaining per-worker output differences outside of agriculture. This may be plausible, as knowledge flows are probably more effective in manufacturing or services than in agriculture. Still, the TFP scales are still tipped against the developing countries. While informative, this exercise does not yet answer the development accounting question of how much PPP income per worker variation would be observed if all countries had the same technology. This counter-factual PPP income per worker would be i i i i i yHK = PA AA yA,KH lA + PA AA yA,KH lA
(16)
where PA and PA are the international-dollar prices of agricultural and non-agricultural goods, and AA and AA are the efficiency levels in some reference country. Because of its additive nature, implementing Equation (16) calls for an explicit choice of values for AA and AA to keep constant as we vary country factor endowments and i i and y i lA l i . (We did not have to sectorial composition – summarized by yA,KH A,KH A choose a benchmark in the purely multiplicative framework of the one-sector model, because in that framework the common level of efficiency disappears when taking the log-variance or the inter-percentile ratio). Furthermore, the results will be sensitive to which country (i.e. what particular choice of As) is chosen as a reference. Hence, the development accountant must decide whose country’s technology will be assigned the role of the benchmark technology in the counterfactual exercise. A certain degree of arbitrariness is inevitable in this choice. Nevertheless, a somewhat plausible argument can be made that it makes sense to hold constant the technology of the richest country in the sample; in our case, the USA. This is because in a sense this is the most successful country, so it is interesting to know how the world income distribution would change if all countries shared the industry TFPs of the most successful among them. For the reference country (the US) it is true by definition that yA = PA AA yA,KH ,
Ch. 9: Accounting for Cross-Country Income Differences
727
and yA = PA AA yA,KH . Since yA and yA are known, we can back out reference values for PA AA and PA AA and compute the counter-factual in (16). The result of this exercise, which is also the main result of this section, is 0.34 for success1 , and 0.32 for success2 . In words, once again, this means that, if all countries had the same industry-level TFPs as the US, but their observed allocation of measurable factors to agriculture and non-agriculture, the world distribution of income would be about one third as unequal as it actually is. Given that – for this sub-sample – the corresponding success measures are 0.45 and 0.39 when the sectorial composition of GDP is not taken into account, I conclude that taking account of differences in sectorial composition actually decreases the share of cross-country income inequality that we can explain with a country’s factor endowments. This result should have been expected, by now. In the previous subsection we have seen that the dispersion in agricultural incomes per worker is a critical “source” of dispersion in per-capita income. Table 5 shows, however, that almost all of the variation in agricultural income comes from differences in agricultural TFP. It is not surprising, therefore, that we find that – even allowing for differences in output composition – factor endowments still do not work as the main cause of GDP differences. One possible way to enhance the quantitative role of sectorial considerations is explored in a highly innovative paper by Graham and Temple (2001). Instead of assuming, as here, that both agriculture and non-agriculture have constant returns to scale, they follow a long tradition in development economics in hypothesizing that the former is characterized by decreasing returns and the latter by increasing returns. As is well known these assumptions tend to generate multiple equilibria, and it is therefore possible to try to explain large cross-country income differences with the argument that poor countries are in “low”, i.e. high agriculture, i.e. low returns, equilibria; while rich countries are in industrialized equilibria and therefore benefit from the increasing returns. The difficulty here is to figure out in the data which countries are in the bad and which ones are in the good equilibrium. The contribution of Graham and Temple is to show a very ingenious way of solving this problem. They find that multiple equilibria explain a relatively large fraction of per capita income differences. The lingering question is whether the significant departures from constant returns to scale required for their result are plausible.64 7. Non-neutral differences in technology 7.1. Basic concepts and qualitative results In all of the previous sections we have assumed that all differences in efficiency across countries are TFP differences, as summarized by the multiplicative factor A. This im64 See also Chanda and Dalgaard (2003) for another contribution that argues for a large role of agriculture.
728
F. Caselli
plies that we view differences in efficiency as factor neutral: some countries simply use all of their inputs more efficiently than others. This is of course a restriction on the set of possible efficiency differences. Caselli and Coleman (2005) have begun exploring a more general view, that allows for the possibility that differences in technology show up as differences in the efficiency with which specific factors – as opposed to all factors proportionally – are used, or even that some countries use some factors more efficiently, and some less efficiently, than others. In other words, a more general view of technology differences where such differences are not factor neutral.65 Extending the development-accounting exercise to allow for factor non neutrality in efficiency differences is the object of this section. While Caselli and Coleman consider a three-factor production function (capital, skilled labor, and unskilled labor), here I will stick to the two-factor world (human and physical capital) of the rest of the chapter. The first step we need to take to proceed in this direction is to replace the Cobb–Douglas restriction – which implicitly rules out non-neutrality – with a more general production function where non-neutral differences can be contemplated. The simplest such generalization is provided by the CES formula: 1/σ Y = α(Ak K)σ + (1 − α)(Ah Lh)σ (17) , α ∈ (0, 1), σ < 1. In (17) Ak and Ah are the efficiency units delivered by one unit of physical capital and one unit of quality-adjusted labor, respectively. If Ak is higher in one country than in another, we say that the former country uses capital more efficiently. If Ah is greater, the country uses human capital more efficiently. The parameters σ and α are constant across countries. σ governs the ease of substitution between physical and human capital. The elasticity of substitution is η = 1/(1 − σ ). The Cobb–Douglas case of the previous sections of the paper emerges as a limit for σ approaching 0 (η approaching 1). In this case, total factor productivity A converges to Aαk Ah1−α .66 In the factor-neutral world explored so far in this chapter, making inference about efficiency differences across countries is a simple matter of solving one equation in one unknown. Inference on non neutral differences is a bit more challenging, as Equation (17) has two unknowns: Ak and Ah . The issue, then, is to find a suitable second equation. As in Caselli and Coleman (2005), to do so I assume that factor markets are everywhere competitive. Then, if r is the user cost of capital, and if w is the market
65 See also Hsieh (2000) for some observations on this topic. 66 Many macroeconomists are attached to the Cobb–Douglas assumption on the alleged ground that the
capital share is constant in the US. The trendlessness of the capital share in the US, however, can of course be replicated by CES models with the “right” time series behavior of the effective supplies of capital and labor (i.e. Ak K and Ah hL). Furthermore, there is clear evidence of substantial fluctuations in the capital shares of many countries other than the US.
Ch. 9: Accounting for Cross-Country Income Differences
729
price of a unit of human capital, the following equations will hold: r = αy 1−σ k σ −1 Aσk , w = (1 − α)y 1−σ hσ −1 Aσh .
(18)
Given values of α and σ , and data on y, k, h, r, and w, these two equations can be solved for the two unknowns Ak and Ah .67 Rearranging Equations (18) we find the following formulas for the factor-specific efficiency levels 1/σ Sk y rk 1 1/σ y = , Ak = y α k α k (19) 1/σ 1/σ Sh y y wh 1 = , Ah = y 1−α k 1−α h where Sk and Sh are the shares of physical and human capital in income, respectively. To see what these equations tell us about the way technology differs across countries it is useful to start from the case where factor shares are constant across countries, i.e. Sk and Sh are invariant parameters. Note that this is the assumption we have maintained so far throughout the paper, where we have set Sk to α (and consequently Sh to 1 − α). Under this assumption, these equations have very intuitive implications: a high output–capital ratio implies that capital is used efficiently, and the same for human capital. Figure 18 plots the output–capital ratio y/k against the log of per-capita income, and Figure 19 does the same for the output–human capital ratio, y/ h. As is well known, the output–capital ratio is decreasing in income. The output–human capital ratio, instead, is increasing. Hence, if we continue to assume that factor shares are constant across countries, but we allow for non-neutrality in technology differences, we reach the startling conclusion that rich countries use human capital more efficiently than poor countries, but they use physical capital less efficiently.68 Consider now relaxing the assumption of constant factor shares. Clearly our conclusions would be unchanged if the factor shares, while not constant, were not systematically related with income. Our knowledge of cross-country patterns in factor shares is somewhat limited. The only thing we are quite sure of is that in the US this share has historically been rather stable, at around 1/3. When it comes to cross-country comparisons, however, we are on shakier ground. Traditionally, the capital share – as measured in the national accounts – is calculated as a residual after employee compensation has been taken out. With this method, Sk is generally found to be higher in poor countries 67 Alternatively, one could solve the system constituted by one of the factor-pricing equations and Equa-
tion (17). The result would be identical. The properties of the constant returns to scale production function, combined with national-account identities, imply that from any two of these equations the third follows. 68 Notice that, in the neutral world of the first part of the chapter, one way of writing total factor productivity is as A = (y/k)α (y/ h)1−α . Hence, our conclusion there that rich countries are more efficient was based on the fact that the increasing pattern of (y/ h)1−α more than compensates for the decreasing pattern in (y/k)α .
730
F. Caselli
Figure 18. Distribution of y/k.
Figure 19. Distribution of y/ h.
Ch. 9: Accounting for Cross-Country Income Differences
731
Figure 20. Distribution of Sk .
than in rich countries. Recently, however, Gollin (2002), and Bernanke and Gurkaynak (2001) have convincingly criticized the construction of the traditional estimates of the capital share, and have provided revised estimates that – among other things – attempt to include the labor component of self-employment income in the labor share. These estimates are plotted in Figure 20.69 Figure 20 shows essentially no systematic pattern of cross-country variation in capital shares. This supports our preliminary finding: the efficiency of capital is higher in poor countries, and the efficiency of (quality adjusted) labor is higher in rich ones! Unfortunately, the data set on capital shares is small – only 54 observations – and developed economies are over-represented. Furthermore, many untested assumptions have been used to develop these estimates. Hence, the conclusion that capital shares are not systematically related to labor productivity is not iron tight. What would it take then to reverse the startling result that poor countries are more efficient users of capital? If factor shares vary systematically with per-worker income, then it becomes critical to know what is the elasticity of substitution η = 1/(1 − σ ). Suppose that Sk is higher in rich countries. If σ > 0 (i.e. η > 1, or capital and human capital are good substitutes
69 The numbers are from Table X in Bernanke and Gurkaynak (2001). We follow their advice and use the
value in column “Actual OSPUE” whenever available; “Imputed OSPUE” when “Actual OSPUE” is unavailable but “Imputed OSPUE” is; and “LF” when the two OSPUE measuresd are unavailable. Of course, Bernanke and Gurkaynak (2001) are reporting labor income, so our measure is 1 minus the numbers in the table.
732
F. Caselli Table 6 Predicted correlations between Ak and y, and Ah and y
σ > 0 (η > 1) σ < 0 (η < 1)
Corr(Sk , y) > 0
Corr(Sk , y) = 0
Corr(Sk , y) < 0
?, ? −, +
−, + −, +
−, + ?, ?
relative to the Cobb–Douglas case), then Ak may conceivably become increasing in income [if (Sk )1/σ grows “faster” than y/k falls]. In this case, however, since Sh = 1−Sk the result on Ah could also possibly be overturned. If σ < 0 (or η < 1) the results from the constant-share case would be reinforced. Symmetrically, if Sk is decreasing in income, the negative (positive) correlation between Ak (Ah ) and y would be reinforced for σ > 0 (η > 1), and weakened (and possibly overturned) if σ < 0 (η < 1). These observations are summarized in Table 6. Each cell of the table lists the predicted sign (positive, negative, or ambiguous) for the correlation between Ak and y (first term) and between Ah and y (second term), conditional on the observed patterns of y/k and y/ h, under various assumptions on σ , and on the correlation between Sk and y. The intuition for the way observed factor shares modify our predictions on crosscountry efficiency patterns is simple. If σ > 0 the two factors are good substitutes. Because the two factors are good substitutes, it makes sense to try to increase the usage of the most efficient factor. Hence, when σ > 0 demand will concentrate on the factor with high efficiency, leading to a high share in income for this factor. Conversely, then, with σ > 0, when we observe a high income share for factor x we can infer that this factor is efficient. On the other hand, if σ < 0 the two factors are poor substitutes. In this case, allocative efficiency calls for boosting the overall efficiency units provided by the low-efficiency factor. This increases the income share of this factor. Hence, with σ < 0, a high income share for factor x signals that this factor is used inefficiently. In sum, skepticism about the greater capital efficiency of poor countries is authorized if one believes that there is a strong positive correlation between Sk and income and η > 1; or if one believes that there is a strong negative correlation between Sk and y and η < 1. We have seen what the data say about Sk (no correlation): what about η? Hamermesh (1993) provides an exhaustive survey, featuring firm, industry, and country-level studies, both cross-sectional and time series. Unfortunately, he reports a dismayingly wide range of estimates, both greater and less than one. To my knowledge, additional recent contributions have not helped narrowing down the region in which η may fall.70 Since published estimates of η are neither stable, nor reliable, one could, perhaps, turn to theoretical considerations. There is of course a tradition of arguing that 70 Aside from the huge dispersion in existing estimates of η, the non-neutrality approach we follow here
points to an intrinsic pitfall in attempting to identifying this parameter. Specifically, empirical investigations of the elasticity of substitution implicitly assume that there is no variation across observations in the relative efficiency of labor and capital. If Ak and Ah vary across observations, then the effective input Ak k and Ah h
Ch. 9: Accounting for Cross-Country Income Differences
733
Table 7 Regressions of log(Ak ) and log(Ah ) on log(y) Dep. var.
η = 0.1
η = 0.5
η = 0.9
η = 1.1
η = 1.5
η=2
η = 50
log(Ak )
−0.32 (6.2) 0.80 (28.06)
−0.027 (4.28) 0.74 (17.55)
0.14 (0.39) 0.21 (.88)
−0.89 (1.99) 1.53 (5.36)
−0.48 (3.58) 1.00 (12.88)
−0.43 (4.38) 0.94 (17.37)
−0.37 (5.59) 0.87 (25.57)
log(As )
long-run elasticities are higher than short-run ones, and macro-economic higher than micro-economic. Ventura (1997) is a particularly convincing recent example. For our purposes it clearly seems appropriate to focus on a long-run, aggregate interpretation of the elasticity. However, it is not clear that even in this case these arguments put a lower bound on η: even accepting that it is higher than a microeconomic, short-run elasticity, does not necessarily imply that it is, say, greater than 1. For the countries with available data, we can actually compute the implied values of Ak and As from Equation (19) for different values of the elasticity of substitution η. For each of these implied set of estimates, in Table 7 we report the coefficients of regressions of log(Ak ) and log(Ah ) on log(y) (t-statistics in parenthesis).71 The results from the table confirm that the available data is mostly consistent with the situation in the middle column of Table 6: for most values of η, rich countries seem to use capital less efficiently than poor ones. The only exception is for η = 0.9, where Ak is weakly and insignificantly positively related to income. This is not surprising, since η = 0.9 is “almost” Cobb–Douglas, and in this limiting case Ak and Ah are not independently identifiable. The coefficients are also sizable, with a 10 percent increase in income per worker being associated with up to a 9 percent decline in Ak , and even larger increases in Ah . Everything considered, the result that poor countries use capital more efficiently than rich ones seems surprisingly robust, particularly because very little structure is imposed on the data to reach this conclusion. Needless to say, it is also rather stunning – especially if one is used to think about the world in TFP (factor neutral) terms. However, a possible theoretical explanation is readily available. will be mis-measured, perhaps wildly. I believe this may indeed be the reason why estimates of η are so unstable. I think this point is implicit in the analysis of Diamond, McFadden and Rodriguez (1978). If the induced measurement error is random, it seems the bias in the estimate of η should be upwards. Intuitively, observations with very different input combinations will appear to have similar output levels, something that is consistent with a high elasticity of substitution. However, if the As vary systematically, the bias could also be downward. Suppose, for example, that Ax and x are positively correlated across observations. Then the data will tend to understate the true variation in effective input, so that less substitutability will appear to be required to explain the observed variation in output. 71 In these regressions there are 53 observations for log(A ) and 50 for log(A ). Also notice that, from k h Equations (19), the As are identified up to the common multiplicative constants α 1/σ and (1 − α)1/σ .
734
F. Caselli
Caselli and Coleman (2005) find somewhat analogous evidence that poor countries use unskilled labor more efficiently than rich countries – while rich countries use skilled labor more efficiently. To explain this finding they develop a simple model of appropriate technology, in which countries face a menu of technology choices. The choice of technology is not neutral, in that different technologies augment different inputs differently. One key result is that which technology is chosen by each country depends on the elasticity of substitution between inputs. If the elasticity of substitution is greater than 1, the appropriate technology augments the abundant factor relatively more, while if the elasticity is less than 1, it is appropriate to choose a technology that augments the scarce factor relatively more. Since the elasticity of substitution between skilled and unskilled workers is commonly deemed to be in the neighborhood of 1.4, and rich countries are abundant in skilled labor, this explains why they would choose skilled-labor augmenting technologies, while poor countries choose unskilled-labor augmenting technologies. Returning now to the new result here that poor countries use physical capital more efficiently – and human capital less efficiently – than rich countries, and recalling that poor countries are relatively abundant in human capital, we can use the same theoretical explanation if we are willing to assume that the elasticity of substitution between capital and (quality-adjusted) labor is less than 1; an assumption that – as we have seen – cannot be ruled out based on the available evidence. In sum, with a high elasticity of substitution between skilled and unskilled labor, and a low elasticity of substitution between capital and the labor aggregate, an appropriate technology model can explain the joint patterns of cross-country choice of the efficiency of capital, skilled labor, and unskilled labor.72 7.2. Development accounting with non-neutral differences Development accounting asks how the observed distribution of GDP per worker compares to the distribution that would obtain in the counterfactual case that all countries had the same technology. As is clear from Equation (17), we are again in the situation in which – unlike in the simple TFP framework – the answer to this question will not be insensitive to which particular pair of values of Ak and Ah are plugged in. As we did in the previous subsection, and for the same reasons, I choose as the benchmark country the USA. Hence, I first compute the US Ak and Ah from Equations (19), and then I plug these numbers in Equation (17) for each country, and obtain measures of success of a model where all countries use the same (US) technology.73
72 Other treatments of the problem studied by Caselli and Coleman (2005) – a country’s appropriate choice
from a menu of technologies as a function of its factor endowments – are, e.g., Basu and Weil (1998) and Acemoglu and Zilibotti (2001). The problem is also closely related to the problem of induced innovation/directed technical change studied, e.g., in Samuelson (1965, 1966) and Acemoglu (1998, 2003). σ 73 More precisely, I compute αAσ = S σ σ k,US (yUS /kUS ) , and (1 − α)Ah = Sh,US (yUS / hUS ) , and plug k these expressions in Equation (17).
Ch. 9: Accounting for Cross-Country Income Differences
735
Figure 21. Success with non-neutral technology differences.
Figure 21 plots the success of the factor-only model when all countries use the technology of the United States. Note that our measures of success converge – as they should – to those of the factor-neutral (Cobb–Douglas) model for η converging to 1. Success of the factor-only model is also systematically decreasing with the value of the elasticity of substitution, η. To see why this is so recall that the lower is η the closer the production function becomes to the Leontief case. Also notice that Figures 18 and 19 imply that poor countries are relatively abundant in human capital: the ratio h/k is decreasing with per-capita income. Finally, recall that the US uses human-capital relatively efficiently. Hence, the ratio Ah,US h/Ak,US k is extremely high in poor countries, suggesting that much “effective human capital” in these countries would be unproductive in the limiting Leontief case. This “waste” may explain the low GDP of poor countries without having to invoke low As, i.e. it leads to greater success of the factor-only model. On the other side of η = 1, we tend to approach the linear production function. The closer we are to this case, the less a disproportionate Ah h/Ak k ratio hurts a country’s productivity, so the factor-only model performs less and less well. Something else is required to explain why GDP is so low. The most remarkable finding of Figure 21, however, is quantitative: namely, not only for elasticities of substitution less than 1 does the model with non neutral technology outperforms the Cobb–Douglas one, but there is a range of elasticities such that the performance of the model is extremely good. Indeed, an elasticity of 0.5 delivers a
736
F. Caselli
perfect fit for the factor-only model! This is remarkable because an elasticity of 0.5 – given our current state of knowledge – is not particularly implausible.74 One could object to the exercise we just reported that the counter-factual choice of technology – all countries using the US technology – is not sensible. For, we know from the previous section that – assuming our factor share data to be dependable – countries are observed to use technologies that cannot be ranked: some technologies boost the productivity of physical capital, some others of human capital. Therefore, one may speculate that – given the choice – poor countries would not necessarily choose to use the US technology. Instead, they would choose the technology most appropriate given their factor endowments. In order to address this point, we now treat the observed (Ak , Ah ) combinations from the previous sub-section as a “menu” of available technologies. One could think of this menu as a summary of the world’s technical knowledge, each observed (Ak , Ah ) pair being a particular blueprint to generate output from physical and human capital. I then re-interpret the counter-factual of no technology differences as one where all countries have access to the same menu of technologies; ask what technology from this menu would each country choose; and compute the counter-factual world income distribution when all countries choose their appropriate technology (from the set of available ones). The appropriate technology is the output-maximizing one.75 The results from this alternative counter-factual experiment are plotted in Figure 22. Here, for each value of η, I have computed from Equations (19) the implied values of Ak and Ah for each country with complete data – including data on the factor shares Sk and Sh . This gave us 50 “observed” technologies. For each country in our 94-country sample, then, I have “chosen” from this menu of 50 the appropriate technology, and I have computed success under the assumption that each country uses this output-maximizing choice. As can be seen, the results preserve the broad qualitative features of those of Figure 21, but quantitatively the factor-only model does much less well. In particular (almost), complete success only occurs if the technology is Leontief. The intuition for this change in results is simple. Given the observed wide disparity in factor proportions, when countries can choose their technology appropriately they will in general choose different (Ak , Ah ) combinations. In particular, few poor countries will find it optimal to use the technology observed in the USA. Indeed, countries with unfavorable factor endowments will be able to partially remedy by choosing technology appropriately. Hence, factor endowments do not have as much explanatory power for income differences as they do when all countries are forced to use the same (US) technology. Even when countries choose technology, however, the measures of success are very sensitive to the choice of the elasticity of substitution: if the elasticity of substitution is 74 Elasticities less than 0.5 imply an income dispersion under the common-technology assumption that are
even greater than the observed ones. 75 See Caselli and Coleman (2005) for a model where the technology choice is decentralized at the level of
firms, and the equilibrium aggregate technology is the GDP-maximizing one.
Ch. 9: Accounting for Cross-Country Income Differences
737
Figure 22. Success with non-neutral but appropriate technology.
low factor endowments can explain a substantially larger share of income differences than in the Cobb–Douglas case (and if it is high a substantially smaller share). It is therefore appropriate to conclude that the Cobb–Douglas assumption is a very sensitive one for development accounting, and that seemingly innocuous generalizations of this assumption – such as the CES formulation employed in this section – can lead to radical changes in results.
8. Conclusions Development accounting is a powerful tool to getting started thinking about the sources of income differences across countries. As of now, the answer to the developmentaccounting question – do observed differences in the factors employed in production explain most of the cross-country variation in income – is: no, way no. This negative answer is robust to attempts to improve the measurement of human capital by allowing for differences in the quality of schooling and in health status of the population; to attempts to account for the age composition of the capital stock; to sectorial disaggregations of output; and to several other robustness checks. On the other hand, incomplete knowledge about certain key parameters that describe the relationship between inputs and outputs implies that the jury should be treated as being still out. For one thing, depending on the elasticity of substitution between capital of different types, the observed wild heterogeneity in the composition of the capital
738
F. Caselli
stock by type of equipment could turn out to be a key proximate determinant of income differences. For another, depending on the elasticity of substitution between physical and human capital, we may find that all is needed for these factors to explain a large fraction of income inequality is a departure from Cobb–Douglas. Disaggregating the government sector out of the data may also potentially reduce the unexplained component of GDP. There is no deep reason why we should not be able to make progress on these three fronts, so that my assessment of the future of this research enterprise is optimistic.
Acknowledgements I thank John Duffy and Silvana Tenreyro for comments, and Mariana Colacelli, Kalina Manova, and Andrea Szabó for research assistance.
References Acemoglu, D. (1998). “Why do new technologies complement skills? Directed technical change and wage inequality”. Quarterly Journal of Economics 113, 1055–1089. Acemoglu, D. (2003). “Labor- and capital-augmenting technical change”. Journal of European Economic Association 1, 1–37. Acemoglu, D., Angrist, J. (2000). “How large are human capital externalities? Evidence from compulsory schooling laws”. In: Bernanke, B.S., Rogoff, K. (Eds.), NBER Macroeconomics Annual 2000. MIT Press, Cambridge, pp. 9–59. Acemoglu, D., Zilibotti, F. (2001). “Productivity differences”. Quarterly Journal of Economics 116, 563–606. Aiyar, S., Dalgaard, C.-J. (2002). “Total factor productivity revisited: A dual approach to levels-accounting”. Working Paper, Brown University. Altonji, J.G., Dunn, T.A. (1996). “The effects of family characteristics on the return to education”. The Review of Economics and Statistics 78 (4), 692–704. Baily, M.N., Solow, R.M. (2001). “International productivity comparisons built from the firm level”. Journal of Economic Perspectives 15 (3), 151–172. Barro, R.J., Lee, J. (2001). “International data on educational attainment: Updates and implications”. Oxford Economic Papers 53 (3), 541–563. Basu, S., Weil, D.N. (1998). “Appropriate technology and growth”. The Quarterly Journal of Economics 113 (4), 1025–1054. Behrman, J.R., Rosenzweig, M.R. (2004). “Returns to birthweight”. Review of Economics and Statistics 86 (2), 586–601. Bernanke, B.S., Gurkaynak, R.S. (2001). “Is growth exogenous? Taking Mankiw, Romer, and Weil seriously”. In: Bernanke, B.S., Rogoff, K. (Eds.), NBER Macroeconomics Annual 2001. MIT Press, Cambridge. Bils, M., Klenow, P.J. (2000). “Does schooling cause growth”. The American Economic Review 90 (5), 1160–1183. Card, D. (1999). “The causal effect of education on earnings”. In: Ashenfelter, O., Card, D. (Eds.), Handbook of Labor Economics, vol. 3A. North-Holland, Amsterdam, pp. 1801–1863. Card, D., Krueger, A. (1996). “Labor market effects of school quality: Theory and evidence”. In: Burtless, G. (Ed.), Does Money Matter? The Effect of School Resources on Student Achievement and Adult Success. Brookings Institute, Wasinghton, DC.
Ch. 9: Accounting for Cross-Country Income Differences
739
Caselli, F., Coleman, W.J. (2001a). “Cross-country technology diffusion: The case of computers”. American Economic Review 91 (2), 328–335. Caselli, F., Coleman, W.J. (2001b). “The U.S. structural transformation and regional convergence: A reinterpretation”. Journal of Political Economy 109 (3), 584–616. Caselli, F., Coleman, W.J. (2005). “The world technology frontier”. Working Paper, LSE. American Economic Review. In press. Caselli, F., Esquivel, G., Lefort, F. (1996). “Reopening the convergence debate: A new look at cross-country growth empirics”. Journal of Economic Growth 1 (3), 363–389. Caselli, F., Tenreyro, S. (2004). “Is Poland the next Spain?”. Working Paper, LSE. Caselli, F., Wilson, D. (2004). “Importing technology”. Journal of Monetary Economics 51 (1), 1–32. Chanda, A., Dalgaard, C.-J. (2003). “Dual economies and international TFP differences”. Working Paper, Louisiana State University. Christensen, L.R., Cummings, D., Jorgenson, D.W. (1981). “Relative productivity levels, 1947–1973: An international comparison”. European Economic Review 16 (1), 61–94. Clark, C. (1940). “The morphology of economic growth”. In: Clark, C. (Ed.), The Conditions of Economic Progress. Macmillan, London, pp. 337–373. Clark, G. (1987). “Why isn’t the whole world developed? Lessons from the cotton mills”. Journal of Economic History 47, 141–173. Conrad, K., Jorgenson, D.W. (1985). “Sectoral productivity gaps between the United States, Japan and Germany, 1960–1979”. In: Giersch, H. (Ed.), Probleme und Perspektiven der weltwirtschaftlichen Entwicklung. Duncker and Humblot, Berlin, pp. 335–347. Currie, J., Thomas, D. (1999). “Early test scores, socioeconomic status and future outcomes”. Working Paper No. 6943, National Bureau of Economic Research. De la Fuente, A., Doménech, R. (2002). “Human capital in growth regressions: How much difference does data quality make? An update and further results”. CEPR Discussion Papers No. 3587. Denison, E.F. (1967). Why Growth Rates Differ. The Brookings Institutions, Washington, DC. Diamond, P., McFadden, D., Rodriguez, M. (1978). “Measurement of the elasticity of factor substitution and bias of technical change”. In: Fuss, M., McFadden, D. (Eds.), Production Economics: A Dual Approach to Theory and Applications, vol. II, Applications of the Theory of Production. North-Holland, Amsterdam. Duflo, E. (2001). “Schooling and labor market consequences of school construction in Indonesia: Evidence from an unusual policy experiment”. American Economic Review 91 (4), 795–813. Easterly, W., Levine, R. (2001). “It’s not factor accumulation: Stylized facts and growth models”. World Bank Economic Review 15 (2), 177–219. Eaton, J., Kortum, S. (2001). “Trade in capital goods”. European Economic Review 45 (7), 1195–1235. Feenstra, R.C. (2000). ‘World trade flows, 1980–1997”. University of California. Feyrer, J. (2002). “Demographics and productivity”. Working Paper, Dartmouth College. Fisher, A. (1945). Economic Programs and Social Security. Macmillan, London. Gelb, A., Knight, J.B., Sabot, R.H. (1988). “Lewis through a looking glass: Public sector employment, rent seeking, and economic growth”. World Bank. Gollin, D. (2002). “Getting income shares right”. Journal of Political Economy 110 (2), 458–474. Gollin, D., Parente, S.L., Rogerson, R. (2000). “Farm work, home work and international productivity differences”. Working Paper. Gollin, D., Parente, S.L., Rogerson, R. (2001). “Structural transformation and cross-country income differences”. Working Paper, University of California in Los Angeles. Gourinchas, P.-O., Jeanne, O. (2003). “The elusive gains from international financial integration”. Working Paper No. 9684, National Bureau of Economic Research. Graham, B.S., Temple, J. (2001). “Rich nations, poor nations: How much can multiple equilibria explain?”. Working Paper No. 76. Center for International Development, Harvard University. Greenwood, J., Hercowitz, Z., Krusell, P. (1997). “Long-run implications of investment-specific technological change”. American Economic Review 87 (3), 342–362. Hall, R.E., Jones, C.I. (1999). “Why do some countries produce so much more output per worker than others?”. The Quarterly Journal of Economics 114 (1), 83–116.
740
F. Caselli
Hamermesh, D.S. (1993). Labor Demand. Princeton University Press. Hanushek, E.A. (2003). “The failure of input based schooling policies”. Economic Journal 113 (485), F64– F98. Hanushek, E.A. (2004). “Some simple analytics of school quality”. Working Paper No. 10229, National Bureau of Economic Research. Hanushek, E.A., Kimko, D.D. (2000). “Schooling, labor-force quality, and the growth of nations”. The American Economic Review 90 (5), 1184–1208. Hanushek, E.A., Rivkin, S.G., Taylor, L.L. (1996). “Aggregation and the estimated effects of school resources”. The Review of Economics and Statistics 78 (4), 611–627. Harrigan, J. (1997). “Technology, factor supplies, and international specialization: estimating the neoclassical model”. American Economic Review 87 (4), 475–494. Harrigan, J. (1999). “Estimation of cross-country differences in aggregate production functions”. Journal of International Economics 47, 267–293. Harris, J.R., Todaro, M. (1970). “Migration, unemployment and development: a two-sector analysis”. American Economic Review 60, 126–142. Heckman, J.J., Klenow, P.J. (1997). “Human capital policy”. Working Paper, University of Chicago. Hendricks, L. (2002). “How important is human capital for development? Evidence from immigrant earnings”. American Economic Review 92 (1), 198–219. Heston, A., Summers, R., Aten, B. (2002). Penn World Tables Version 6.1. Downloadable Dataset. Center for International Comparisons at the University of Pennsylvania. Hill, R. (2000). “Measuring substitution bias in international comparisons based on additive purchasing power parity methods”. European Economic Review 44 (1), 45–162. Hsieh, C.-T. (2000). “Measuring biased technology”. Working Paper, Princeton University. Hulten, C.R. (1996). “Infrastructure capital and economic growth: How well you use it may be more important than how much you have”. Working Paper No. 5847, National Bureau of Economic Research. Hsieh, C.-T., Klenow, P.J. (2003). “Relative prices and relative prosperity”. Working Paper No. 9701, National Bureau of Economic Research. Islam, N. (1995). “Growth empirics: A panel data approach”. The Quarterly Journal of Economics 110, 1127– 1170. Johnson, G.E., Stafford, F.P. (1973). “Social returns to quantity and quality of schooling”. Journal of Human Resources 8 (2), 139–155. Jorgenson, D.W. (1961). “The development of a dual economy”. Economic Journal 71, 309–334. Jorgenson, D.W. (1995a). Productivity, vol. 1: Postwar U.S. Economic Growth. MIT Press. Jorgenson, D.W. (1995b). Productivity, vol. 2: International Comparisons of Economic Growth. MIT Press. Jorgenson, D.W., Gollop, F.M. (1992). “Productivity growth in U.S. agriculture: A postwar perspective”. American Journal of Agricultural Economics 74 (3), 745–750. Jovanovic, B., Rob, R. (1997). “Solow v. Solow: Machine prices and development”. Unpublished, New York University. King, R.G., Levine, R. (1994). “Capital fundamentalism, economic development, and economic growth”. Carnegie-Rochester Conference Series on Public Policy 40 (0), 259–292. Klenow, P.J., Rodriguez-Clare, A. (1997). “The neoclassical revival in growth economics: Has it gone too far?”. In: Bernanke, B.S., Rotemberg, J.J. (Eds.), NBER Macroeconomics Annual 1997. MIT Press, Cambridge, pp. 73–103. Knight, M., Loayza, N., Villanueva, D. (1993). “Testing the neoclassical theory of economic growth”. IMF Staff Papers 40 (3), 512–541. Koren, M., Tenreyro, S. (2004). “Diversification and development”. LSF Working Paper, 3. Krueger, A.B. (2003). “Economic considerations and class size”. Economic Journal 113 (485), F34–F63. Kuznets, S. (1966). Modern Economic Growth. Yale University Press, New Haven. Lee, J.-W., Barro, R.J. (2001). “Schooling quality in a cross-section of countries”. Economica 68, 465–488. Lewis, W.A. (1954). “Economic development with unlimited supplies of labor”. Manchester School of Economics and Social Studies 22, 139–151.
Ch. 9: Accounting for Cross-Country Income Differences
741
Mankiw, N.G., Romer, D., Weil, D.N. (1992). “A contribution to the empirics of economic growth”. The Quarterly Journal of Economics 107 (2), 407–437. Mateos-Planas, X. (2000). “Schooling and distortions in a vintage capital model”. University of Southampton, Department of Economics. 40 pp. McGrattan, E.R., Schmitz, J.A. Jr. (1999). “Explaining cross-country income differences”. In: Taylor, J.B., Woodford, M. (Eds.), Handbook of Macroeconomics, vol. 1A. North-Holland, Amsterdam, pp. 669–737. Murnane, R.J., Willett, J.B., Levy, F. (1995). “The growing importance of cognitive skills in wage determination”. The Review of Economics and Statistics 77 (2), 251–266. Murphy, K., Shleifer, A., Vishny, R. (1991). “The allocation of talent: Implications for growth”. Quarterly Journal of Economics 106 (2), 503–530. Neal, D.A., Johnson, W.R. (1996). “The role of premarket factors in black–white wage differences”. Journal of Political Economy 104 (5), 869–895. Nurkse, R. (1953). Problems of Capital Formation in Underdeveloped Countries. Oxford University Press, New York. Parente, S.L. (2000). “Learning-by-using and the switch to better machines”. Review of Economic Dynamics 3 (4), 675–703. Prescott, E.C. (1998). “Needed: A theory of total factor productivity”. International Economic Review 39 (3), 525–551. Pritchett, L. (2000). “The tyranny of concepts: CUDIE (Cumulated, Depreciated, Investment Effort) is not capital”. Journal of Economic Growth 5 (4), 361–384. Pritchett, L. (2003). “Does learning to add up add up? The returns to schooling in aggregate data”. Draft chapter for Handbook of Education Economics, Harvard University. Psacharopoulos, G. (1994). “Returns to investment in education: A global update”. World Development 22 (9), 1325–1343. Rao, P.D.S. (1993). “Intercountry comparisons of agricultural output and productivity”. FAO Economic and Social Development Paper, FAO, Rome. Restuccia, D., Yang, D.T., Zhu, X. (2003). “Agriculture and aggregate productivity: A quantitative crosscountry analysis”. Working Paper, University of Toronto, Economics Department. Rodriguez-Clare, A. (1996). “The role of trade in technology diffusion”. Minneapolis Fed Working Paper 114. Rostow, W.W. (1960). The Stages of Economic Growth: A Non-Communist Manifesto. Cambridge University Press, Cambridge, MA. Samuelson, P. (1965). “A theory of induced innovation along Kennedy–Weiszacker lines”. Review of Economics and Statistics 47 (November). Samuelson, P. (1966). “Rejoinder: Agreements, disagreements, doubts, and the case of induced Harrod-neutral technical change”. Review of Economics and Statistics 48 (November). Schmitz, J.A., Jr. (2001). “What determines labor productivity?: Lessons from the dramatic recovery of the U.S. Canadian iron-ore industries”. Staff Report, 286, Federal Reserve Bank of Minneapolis. Shastry, G.K., Weil, D.N. (2003). “How much of cross-country income variation is explained by health?”. Journal of the European Economic Association 1 (2–3), 387–396. Solow, R. (1959). “Investment and technical progress”. In: Arrow, K., Karlin, S., Suppes, P. (Eds.), Mathematical Methods in the Social Sciences. Stanford University Press. Summers, R., Heston, A. (1991). “The Penn World Table (Mark 5): An expanded set of international comparisons, 1950–1988”. The Quarterly Journal of Economics 101 (2), 327–368. Temple, J. (2003). “Dualism and aggregate productivity”. Unpublished, University of Bristol. Ventura, J. (1997). “Growth and interdependence”. The Quarterly Journal of Economics 112 (1), 57–84. Wagner, K., van Ark, B. (Eds.) (1996). International Productivity Differences: Measurement and Explanations. North-Holland, Amsterdam. Weil, D.N. (2001). “Accounting for the effect of health on economic growth”. Working Paper, Brown University. Wilson, D.J. (2004). “Cross-country differences in capital composition”. Forthcoming in FRBSF Economic Letter.
Chapter 10
ACCOUNTING FOR GROWTH IN THE INFORMATION AGE DALE W. JORGENSON Department of Economics, Harvard University, 122 Littauer Center, Cambridge, MA 02138-3001, USA
Contents Abstract Keywords 1. The information age 1.1. Introduction 1.2. Faster, better, cheaper 1.2.1. Moore’s Law 1.2.2. Semiconductor prices 1.2.3. Constant Quality Price Indexes 1.2.4. Computers 1.2.5. Communications equipment and software 1.2.6. Research opportunities 1.3. Impact of information technology
2. Aggregate growth accounting 2.1. The role of information technology 2.1.1. Output 2.1.2. Capital services 2.1.3. Labor services 2.2. The American growth resurgence 2.2.1. Production possibility frontier 2.2.2. Sources of growth 2.2.3. Contributions of IT output 2.2.4. Total factor productivity 2.2.5. Output growth 2.2.6. Average labor productivity 2.2.7. Research opportunities 2.3. Demise of traditional growth accounting 2.3.1. Introduction 2.3.2. Human capital 2.3.3. Solow’s surprise 2.3.4. Radical departure
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01010-5
744 745 746 746 747 748 748 749 750 751 754 755 756 756 757 759 761 765 767 768 770 774 776 778 778 779 779 780 781 782
744
D.W. Jorgenson 2.3.5. The Rees Report
3. International comparisons 3.1. Introduction 3.2. Investment and total factor productivity 3.2.1. Comparisons of output, input, and productivity 3.2.2. Comparisons of capital and labor quality 3.2.3. The relative importance of investment and total factor productivity 3.3. Investment in information technology 3.4. Alternative approaches 3.4.1. Comparisons without growth accounts 3.4.2. Convergence 3.4.3. Modeling productivity differences 3.5. Conclusions
4. Economics on internet time Acknowledgements References Further reading
783 784 784 787 787 791 796 796 803 803 804 805 805 806 807 808 815
Abstract The “killer application” of the new framework for productivity measurement presented in this paper is the impact of information technology (IT) on economic growth. A consensus has emerged that the remarkable behavior of IT prices provides the key to the surge in U.S. economic growth after 1995. The relentless decline in the prices of information technology equipment and software has steadily enhanced the role of IT investment. Productivity growth in IT-producing industries has risen in importance and a productivity revival is underway in the rest of the economy. The surge of IT investment in the United States after 1995 has counterparts in all other industrialized countries. It is essential to use comparable data and methodology in order to provide rigorous international comparisons. A crucial role is played by measurements of IT prices. The U.S. national accounts have incorporated measures of IT prices that hold performance constant since 1985. Schreyer [Schreyer, Paul (2000), “The contribution of information and communication technology to output growth: A study of the G7 Countries”. Organization for Economic Co-operation and Development, Paris, May 23] has extended these measures to other industrialized countries by constructing “internationally harmonized prices”. The acceleration in the IT price decline in 1995 triggered a burst of IT investment in all of the G7 nations – Canada, France, Germany, Italy, Japan, the UK, as well as the U.S. These countries also experienced a rise in productivity growth in the ITproducing industries. However, differences in the relative importance of these industries have generated wide disparities in the impact of IT on economic growth. The role of the IT-producing industries is greatest in the U.S., which leads the G7 in output per capita.
Ch. 10:
Accounting for Growth in the Information Age
Keywords productivity, growth, information, technology, investment JEL classification: C82, D24, E23
745
746
D.W. Jorgenson
1. The information age 1.1. Introduction The resurgence of the American economy since 1995 has outrun all but the most optimistic expectations. Economic forecasting models have been seriously off track and growth projections have been revised repeatedly to reflect a more sanguine outlook.1 It is not surprising that the unusual combination of more rapid growth and slower inflation touched off a strenuous debate about whether improvements in America’s economic performance could be sustained. The starting point for the economic debate is the thesis that the 1990s are a mirror image of the 1970s, when an unfavorable series of “supply shocks” led to stagflation – slower growth and higher inflation.2 In this view, the development of information technology (IT) is one of a series of positive, but temporary, shocks. The competing perspective is that IT has produced a fundamental change in the U.S. economy, leading to a permanent improvement in growth prospects.3 The resolution of this debate in favor of a permanent improvement has been the “killer application” of a new framework for productivity measurement summarized in Paul Schreyer’s (2001) OECD Manual. A consensus has emerged that the development and deployment of information technology is the foundation of the American growth resurgence. A mantra of the “new economy” – faster, better, cheaper – captures the speed of technological change and product improvement in semiconductors and the precipitous and continuing fall in semiconductor prices. The price decline has been transmitted to the prices of products that rely heavily on semiconductor technology, like computers and telecommunications equipment. This technology has also helped to reduce the cost of aircraft, automobiles, scientific instruments, and a host of other products. Swiftly falling IT prices provide powerful economic incentives for the substitution of IT equipment for other forms of capital and for labor services. The rate of the IT price decline is a key component of the cost of capital, required for assessing the impacts of rapidly growing stocks of computers, communications equipment, and software. Constant quality price indexes are essential for identifying the change in price for a given level of performance. Accurate and timely computer prices have been part of the U.S. National Income and Product Accounts (NIPA) since 1985. Unfortunately, important information gaps remain, especially on trends in prices for closely related investments, such as software and communications equipment. Capital input has been the most important source of U.S. economic growth throughout the postwar period. More rapid substitution toward information technology has
1 See Congressional Budget Office (2002) on official forecasts and Economics and Statistics Administration
(2000, p. 60), on private forecasts. 2 Robert Gordon (1998, 2000); Barry Bosworth and Jack Triplett (2000). 3 Alan Greenspan (2000).
Ch. 10:
Accounting for Growth in the Information Age
747
given much additional weight to components of capital input with higher marginal products. The vaulting contribution of capital input since 1995 has boosted growth by close to a percentage point. The contribution of investment in IT accounts for more than half of this increase. Computers have been the predominant impetus to faster growth, but communications equipment and software have made important contributions as well. The accelerated information technology price decline signals faster productivity growth in IT-producing industries. In fact, these industries have been a rapidly rising source of aggregate productivity growth throughout the 1990s. The IT-producing industries generate less than five percent of gross domestic income, but have accounted for nearly half the surge in productivity growth since 1995. However, it is important to emphasize that faster productivity growth is not limited to these industries. The dramatic effects of information technology on capital and labor markets have already generated a substantial and growing economic literature, but many important issues remain to be resolved. For capital markets the relationship between equity valuations and growth prospects merits much further study. For labor markets more research is needed on investment in information technology and substitution among different types of labor. 1.2. Faster, better, cheaper Modern information technology begins with the invention of the transistor, a semiconductor device that acts as an electrical switch and encodes information in binary form. A binary digit or bit takes the values zero and one, corresponding to the off and on positions of a switch. The first transistor, made of the semiconductor germanium, was constructed at Bell Labs in 1947 and won the Nobel Prize in Physics in 1956 for the inventors – John Bardeen, Walter Brattain, and William Shockley.4 The next major milestone in information technology was the co-invention of the integrated circuit by Jack Kilby of Texas Instruments in 1958 and Robert Noyce of Fairchild Semiconductor in 1959. An integrated circuit consists of many, even millions, of transistors that store and manipulate data in binary form. Integrated circuits were originally developed for data storage and retrieval and semiconductor storage devices became known as memory chips.5 The first patent for the integrated circuit was granted to Noyce. This resulted in a decade of litigation over the intellectual property rights. The litigation and its outcome demonstrate the critical importance of intellectual property in the development of information technology. Kilby was awarded the Nobel Prize in Physics in 2000 for discovery of the integrated circuit; regrettably, Noyce died in 1990.6
4 On Bardeen, Brattain, and Shockley, see: http://www.nobel.se/physics/laureates/1956/. 5 Charles Petzold (1999) provides a general reference on computers and software. 6 On Kilby, see: http://www.nobel.se/physics/laureates/2000/. On Noyce, see: Tom Wolfe (2000, pp. 17–65).
748
D.W. Jorgenson
1.2.1. Moore’s Law In 1965 Gordon Moore, then Research Director at Fairchild Semiconductor, made a prescient observation, later known as Moore’s Law.7 Plotting data on memory chips, he observed that each new chip contained roughly twice as many transistors as the previous chip and was released within 18–24 months of its predecessor. This implied exponential growth of chip capacity at 35–45 percent per year! Moore’s prediction, made in the infancy of the semiconductor industry, has tracked chip capacity for thirty-five years. He recently extrapolated this trend for at least another decade.8 In 1968 Moore and Noyce founded Intel Corporation to speed the commercialization of memory chips.9 Integrated circuits gave rise to microprocessors with functions that can be programmed by software, known as logic chips. Intel’s first general purpose microprocessor was developed for a calculator produced by Busicom, a Japanese firm. Intel retained the intellectual property rights and released the device commercially in 1971. The rapidly rising trends in the capacity of microprocessors and storage devices illustrate the exponential growth predicted by Moore’s Law. The first logic chip in 1971 had 2,300 transistors, while the Pentium 4 released on November 20, 2000, had 42 million! Over this twenty-nine year period the number of transistors increased by thirty-four percent per year. The rate of productivity growth for the U.S. economy during this period was slower by two orders of magnitude. 1.2.2. Semiconductor prices Moore’s Law captures the fact that successive generations of semiconductors are faster and better. The economics of semiconductors begins with the closely related observation that semiconductors have become cheaper at a truly staggering rate! Figure 1 gives semiconductor price indexes constructed by Bruce Grimm (1998) of the Bureau of Economic Analysis (BEA) and employed in the U.S. National Income and Product Accounts since 1996. These are divided between memory chips and logic chips. The underlying detail includes seven types of memory chips and two types of logic chips. Between 1974 and 1996 prices of memory chips decreased by a factor of 27,270 times or at 40.9 percent per year, while the implicit deflator for the gross domestic product (GDP) increased by almost 2.7 times or 4.6 percent per year! Prices of logic chips, available for the shorter period 1985 to 1996, decreased by a factor of 1,938 or 54.1 percent per year, while the GDP deflator increased by 1.3 times or 2.6 percent per year! Semiconductor price declines closely parallel Moore’s Law on the growth of chip capacity, setting semiconductors apart from other products. 7 Moore (1965). Vernon Ruttan (2001, pp. 316–367), provides a general reference on the economics of
semiconductors and computers. On semiconductor technology, see: http://euler.berkeley.edu/~esrc/csm. 8 Moore (2003). 9 Moore (1997).
Ch. 10:
Accounting for Growth in the Information Age
749
Figure 1. Relative prices of computers and semiconductors, 1959–2002. Note: All price indexes are divided by the output price index.
Figure 1 also reveals a sharp acceleration in the decline of semiconductor prices in 1994 and 1995. The microprocessor price decline leaped to more than ninety percent per year as the semiconductor industry shifted from a three-year product cycle to a greatly accelerated two-year cycle. This is reflected in the 2003 Edition of the International Technology Road Map for Semiconductors,10 prepared by a consortium of industry associations. Ana Aizcorbe, Stephen Oliner and Daniel Sichel (2003) have identified and analyzed break points in prices of microprocessors and storage devices. 1.2.3. Constant Quality Price Indexes The behavior of semiconductor prices is a severe test for the methods used in the official price statistics. The challenge is to separate observed price changes between changes in semiconductor performance and changes in price that hold performance constant. Achieving this objective has required a detailed understanding of the technology, the development of sophisticated measurement techniques, and the introduction of novel methods for assembling the requisite information. Ellen Dulberger (1993) introduced a “matched model” index for semiconductor prices. A matched model index combines price relatives for products with the same performance at different points of time. Dulberger presented constant quality price indexes based on index number formulas, including the Fisher (1922) ideal index used 10 On International Technology Roadmap for Semiconductors (2004), see: http://public.itrs.net/.
750
D.W. Jorgenson
in the in the U.S. national accounts.11 The Fisher index is the geometric average of the familiar Laspeyres and Paasche indexes. Erwin Diewert (1976) defined a superlative index number as an index that exactly replicates a flexible representation of the underlying technology (or preferences). A flexible representation provides a second-order approximation to an arbitrary technology (or preference system). A.A. Konus and S.S. Byushgens (1926) first showed that the Fisher ideal index is superlative in this sense. Laspeyres and Paasche indexes are not superlative and fail to capture substitutions among products in response to price changes accurately. Grimm (1998) combined matched model techniques with hedonic methods, based on an econometric model of semiconductor prices at different points of time. A hedonic model gives the price of a semiconductor product as a function of the characteristics that determine performance, such as speed of processing and storage capacity. A constant quality price index isolates the price change by holding these characteristics of semiconductors fixed.12 Beginning in 1997, the Bureau of Labor Statistics (BLS) incorporated a matched model price index for semiconductors into the Producer Price Index (PPI) and since then the national accounts have relied on data from the PPI. Reflecting long-standing BLS policy, historical data were not revised backward. Semiconductor prices reported in the PPI prior to 1997 do not hold quality constant, failing to capture the rapid semiconductor price decline and the acceleration in 1995. 1.2.4. Computers The introduction of the Personal Computer (PC) by IBM in 1981 was a watershed event in the deployment of information technology. The sale of Intel’s 8086-8088 microprocessor to IBM in 1978 for incorporation into the PC was a major business breakthrough for Intel.13 In 1981 IBM licensed the MS-DOS operating system from the Microsoft Corporation, founded by Bill Gates and Paul Allen in 1975. The PC established an Intel/Microsoft relationship that has continued up to the present. In 1985 Microsoft released the first version of Windows, its signature operating system for the PC, giving rise to the Wintel (Windows-Intel) nomenclature for this ongoing collaboration. Mainframe computers, as well as PCs, have come to rely heavily on logic chips for central processing and memory chips for main memory. However, semiconductors account for less than half of computer costs and computer prices have fallen much less rapidly than semiconductor prices. Precise measures of computer prices that hold product quality constant were introduced into the NIPA in 1985 and the PPI during the 11 See Steven Landefeld and Robert Parker (1997). 12 Triplett (2003) has drafted a manual for the OECD on constructing constant quality price indexes for
information technology and communications equipment and software. 13 See Moore (1996).
Ch. 10:
Accounting for Growth in the Information Age
751
1990s. The national accounts now rely on PPI data, but historical data on computers from the PPI, like the PPI data on semiconductors, do not hold quality constant. Gregory Chow (1967) pioneered the use of hedonic techniques for constructing a constant quality index of computer prices in research conducted at IBM. Chow documented price declines at more than twenty percent per year during 1960–1965, providing an initial glimpse of the remarkable behavior of computer prices. In 1985 the Bureau of Economic Analysis incorporated constant quality price indexes for computers and peripheral equipment constructed by IBM into the NIPA. Jack Triplett’s (1986) discussion of the economic interpretation of these indexes brought the rapid decline of computer prices to the attention of a very broad audience. The BEA–IBM constant quality price index for computers provoked a heated exchange between BEA and Edward Denison (1989), one of the founders of national accounting methodology in the 1950s and head of the national accounts at BEA from 1979 to 1982. Denison sharply attacked the BEA–IBM methodology and argued vigorously against the introduction of constant quality price indexes into the national accounts.14 Allan Young (1989), then Director of BEA, reiterated BEA’s rationale for introducing constant quality price indexes. Dulberger (1989) presented a more detailed report on her research on the prices of computer processors for the BEA–IBM project. Speed of processing and main memory played central roles in her model. Triplett (1989, 2003) has provided exhaustive surveys of research on hedonic price indexes for computers. Gordon (1989, 1990) gave an alternative model of computer prices and identified computers and communications equipment, along with commercial aircraft, as assets with the highest rates of price decline. Figure 2 gives BEA’s constant quality index of prices of computers and peripheral equipment and its components, including mainframes, PC’s, storage devices, other peripheral equipment, and terminals [Bureau of Economic Analysis (1995)]. The decline in computer prices follows the behavior of semiconductor prices presented in Figure 1, but in much attenuated form. The 1995 acceleration in the computer price decline parallels the acceleration in the semiconductor price decline that resulted from the changeover from a three-year product cycle to a two-year cycle in 1995. 1.2.5. Communications equipment and software Communications technology is crucial for the rapid development and diffusion of the Internet, perhaps the most striking manifestation of information technology in the American economy.15 Kenneth Flamm (1989) was the first to compare the behavior of computer prices and the prices of communications equipment. He concluded that the 14 Denison cited his 1957 paper, “Theoretical aspects of quality change, capital consumption, and net capital
formation”, as the definitive statement of the traditional BEA position. 15 General references on the economics of the Internet are Choi Soon-Yong and Andrew Whinston (2000)
and Robert Hall (2002). On Internet indicators see: http://www.internetindicators.com/.
752
D.W. Jorgenson
Figure 2. Relative prices of computers, communications, and software, 1948–2002. Note: All price indexes are divided by the output price index.
communications equipment prices fell only a little more slowly than computer prices. Gordon (1990) compared Flamm’s results with the official price indexes, revealing substantial bias in the official indexes. Communications equipment is an important market for semiconductors, but constant quality price indexes cover only a portion of this equipment. Switching and terminal equipment rely heavily on semiconductor technology, so that product development reflects improvements in semiconductors. Grimm’s (1997) constant quality price index for digital telephone switching equipment, given in Figure 3, was incorporated into the national accounts in 1996. The output of communications services in the NIPA also incorporates a constant quality price index for cellular phones. Much communications investment takes the form of the transmission gear, connecting data, voice, and video terminals to switching equipment. Technologies such as fiber optics, microwave broadcasting, and communications satellites have progressed at rates that outrun even the dramatic pace of semiconductor development. An example is dense wavelength division multiplexing (DWDM), a technology that sends multiple signals over an optical fiber simultaneously. Installation of DWDM equipment, beginning in 1997, has doubled the transmission capacity of fiber optic cables every 6–12 months.16 Mark Doms (2004) has provided comprehensive price indexes for terminals, switching gear, and transmission equipment. These have been incorporated into the Federal Reserve’s Index of Industrial Production, as described by Carol Corrado (2003), but are 16 Rick Rashad (2000) characterizes this as the “demise” of Moore’s Law. Jeff Hecht (1999) describes
DWDM technology and provides a general reference on fiber optics.
Ch. 10:
Accounting for Growth in the Information Age
753
Figure 3. Relative prices of computers, central office switching equipment, and prepackaged software, 1959–2002. Note: All price indexes are divided by the output price index.
Figure 4. Output shares of information technology by type, 1948–2002. Note: Share of current dollar gross domestic product.
754
D.W. Jorgenson
not yet included in the U.S. National Income and Product Accounts. The analysis of the impact of information technology on the U.S. economy described below is based on the national accounts and remains incomplete. Both software and hardware are essential for information technology and this is reflected in the large volume of software expenditures. The eleventh comprehensive revision of the national accounts, released by BEA on October 27, 1999, re-classified computer software as investment [Bureau of Economic Analysis (1999)].17 Before this important advance, business expenditures on software were treated as current outlays, while personal and government expenditures were treated as purchases of nondurable goods. Software investment is growing rapidly and is now much more important than investment in computer hardware. Parker and Grimm (2000) describe the new estimates of investment in software. BEA distinguishes among three types of software – prepackaged, custom, and own-account software. Prepackaged software is sold or licensed in standardized form and is delivered in packages or electronic files downloaded from the Internet. Custom software is tailored to the specific application of the user and is delivered along with analysis, design, and programming services required for customization. Own-account software consists of software created for a specific application. However, only price indexes for prepackaged software hold performance constant. Parker and Grimm (2000) present a constant quality price index for prepackaged software, given in Figure 3. This combines a hedonic model of prices for business applications software and a matched model index for spreadsheet and word processing programs developed by Oliner and Sichel (1994). Prepackaged software prices decline at more than ten percent per year over the period 1962–1998. Since 1998 the BEA has relied on a matched model price index for all prepackaged software from the PPI; prior to 1998 the PPI data do not hold quality constant. BEA’s prices for own-account and custom software are based on programmer wage rates. This implicitly assumes no change in the productivity of computer programmers, even with growing investment in hardware and software to support the creation of new software. Custom and own-account software prices are a weighted average of prepackaged software prices and programmer wage rates with arbitrary weights of 75 percent for programmer wage rates and 25 percent for prepackaged software. These price indexes do not hold the software performance constant and present a distorted picture of software prices, as well as software output and investment. 1.2.6. Research opportunities The official price indexes for computers and semiconductors provide the paradigm for economic measurement. These indexes capture the steady decline in IT prices and the
17 Brent Moulton (2000) describes the 11th comprehensive revision of NIPA and the 1999 update.
Ch. 10:
Accounting for Growth in the Information Age
755
recent acceleration in this decline. The official price indexes for central office switching equipment and prepackaged software also hold quality constant. BEA and BLS, the leading statistical agencies in price research, have carried out much of the best work in this area. However, a critical role has been played by price research at IBM, long the dominant firm in information technology.18 It is important to emphasize that information technology is not limited to applications of semiconductors. Switching and terminal equipment for voice, data, and video communications have come to rely on semiconductor technology and the empirical evidence on prices of this equipment reflects this fact. Transmission gear employs technologies with rates of progress that far outstrip those of semiconductors. This important gap in our official price statistics has been filled by constant quality price indexes for all types of communications equipment constructed by Doms (2004), but these indexes have not been incorporated into the national accounts. Investment in software is more important than investment in hardware. This was essentially invisible until BEA introduced new measures of prepackaged, custom, and own-account software investment into the national accounts in 1999. This is a crucial step in understanding the role of information technology in the American economy. Unfortunately, software prices are a statistical blind spot with only prices of prepackaged software adequately represented in the official system of price statistics. The daunting challenge that lies ahead is to construct constant quality price indexes for custom and own-account software. 1.3. Impact of information technology In Section 2 I consider the “killer application” of the new framework for productivity measurement – the impact of information technology on economic growth. Despite differences in methodology and data sources, a consensus has emerged that the remarkable behavior of IT prices provides the key to the surge in U.S. economic growth after 1995. The relentless decline in the prices of information technology equipment and software has steadily enhanced the role of IT investment. Productivity growth in IT-producing industries has risen in importance and a productivity revival is underway in the rest of the economy. A substantial acceleration in the IT price decline occurred in 1995, triggered by a much sharper acceleration in the price decline of semiconductors, the key component of modern information technology. Although the decline in semiconductor prices has been projected to continue for at least another decade, the recent acceleration may be temporary. This can be traced to a shift in the product cycle for semiconductors from three years to two years as a consequence of intensifying competition in markets for semiconductor products. In Section 3 I show that the surge of IT investment in the United States after 1995 has counterparts in all other industrialized countries. It is essential to use comparable data 18 See Alfred Chandler (2000, Table 1.1, p. 26).
756
D.W. Jorgenson
and methodology in order to provide rigorous international comparisons. A crucial role is played by measurements of IT prices. The U.S. national accounts have incorporated measures of IT prices that hold performance constant since 1985. Schreyer (2000) has extended these measures to other industrialized countries by constructing “internationally harmonized prices”.19 I show that the acceleration in the IT price decline in 1995 triggered a burst of IT investment in all of the G7 nations – Canada, France, Germany, Italy, Japan, the United Kingdom, as well as the United States. These countries also experienced a rise in productivity growth in the IT-producing industries. However, differences in the relative importance of these industries have generated wide disparities in the impact of IT on economic growth. The role of the IT-producing industries is greatest in the United States, which leads the G7 in output per capita. Section 4 concludes.
2. Aggregate growth accounting 2.1. The role of information technology At the aggregate level information technology is identified with the outputs of computers, communications equipment, and software. These products appear in the GDP as investments by businesses, households, and governments along with net exports to the rest of the world. The GDP also includes the services of IT products consumed by households and governments. A methodology for analyzing economic growth must capture the substitution of IT outputs for other outputs of goods and services. While semiconductor technology is the driving force behind the spread of IT, the impact of the relentless decline in semiconductor prices is transmitted through falling IT prices. Only net exports of semiconductors, defined as the difference between U.S. exports to the rest of the world and U.S. imports appear in the GDP. Sales of semiconductors to domestic manufacturers of IT products are precisely offset by purchases of semiconductors and are excluded from the GDP. Constant quality price indexes, like those reviewed in the previous section, are a key component of the methodology for analyzing the American growth resurgence. Computer prices were incorporated into the NIPA in 1985 and are now part of the PPI as well. Much more recently, semiconductor prices have been included in the NIPA and the PPI. The official price indexes for communications equipment do not yet reflect the important work of Doms (2004). Unfortunately, evidence on the price of software is seriously incomplete, so that the official price indexes are seriously misleading.
19 The measurement gap in IT prices between the U.S. and other OECD countries was first identified by
Andrew Wyckoff (1995).
Ch. 10:
Accounting for Growth in the Information Age
757
2.1.1. Output The output data in Table 1 are based on the most recent benchmark revision of the national accounts through 2000.20 The output concept is similar, but not identical, to the concept of gross domestic product used by the BEA. Both measures include final outputs purchased by businesses, governments, households, and the rest of the world. Unlike the BEA concept, the output measure in Table 1 also includes imputations for the service flows from durable goods, including IT products, employed in the household and government sectors. The imputations for services of IT equipment are based on the cost of capital for IT described in more detail below. The cost of capital is multiplied by the nominal value of IT capital stock to obtain the imputed service flow from IT products. In the business sector this accrues as capital income to the firms that employ these products as inputs. In the household and government sectors the flow of capital income must be imputed. This same type of imputation is used for housing in the NIPA. The rental value of renteroccupied housing accrues to real estate firms as capital income, while the rental value of owner-occupied housing is imputed to households. Current dollar GDP in Table 1 is $11.3 trillions in 2002, including imputations, and real output growth averaged 3.46 percent for the period 1948–2002. These magnitudes can be compared to the current dollar value of $10.5 trillions in 2002 and the average real growth rate of 3.36 percent for period 19480–2002 for the official GDP. Table 1 presents the current dollar value and price indexes of the GDP and IT output. This includes outputs of investment goods in the form of computers, software, communications equipment, and non-IT investment goods. It also includes outputs of non-IT consumption goods and services as well as imputed IT capital service flows from households and governments. The most striking feature of the data in Table 1 is the rapid price decline for computer investment, 15.8 percent per year from 1959 to 1995. Since 1995 this decline has increased to 31.0 percent per year. By contrast the relative price of software has been flat for much of the period and began to fall only in the 1980s. The price of communications equipment behaves similarly to the software price, while the consumption of capital services from computers and software by households and governments shows price declines similar to computer investment. The top panel of Table 2 summarizes the growth rates of prices and quantities for major output categories for 1989–1995 and 1995–2002. Business investments in computers, software, and communications equipment are the largest categories of IT spending. Households and governments have also spent sizable amounts on computers, software and communications equipment. Figure 1 shows that the share of software output in the GDP is largest, followed by the shares of communications equipment and computers.
20 See Jorgenson and Stiroh (2000b, Appendix A), for details on the estimates of output.
758
D.W. Jorgenson Table 1 Information technology output and gross domestic product
Computer Year 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990
Value
0.0 0.2 0.3 0.3 0.8 1.0 1.3 1.9 2.1 2.1 2.7 3.0 3.2 4.1 4.2 4.8 4.6 5.6 7.2 9.7 13.2 17.3 22.6 25.4 34.8 43.4 46.0 45.7 48.6 54.0 56.8 52.3
Price
1,635.06 1,635.06 1,226.29 817.53 572.27 490.52 408.76 283.63 228.43 194.16 176.76 158.80 142.57 128.27 142.84 128.85 152.48 125.12 98.49 60.49 45.22 34.18 25.95 25.83 20.42 18.70 15.41 13.64 12.40 12.16 12.01 10.86
Software Value
0.1 0.2 0.2 0.5 0.6 0.8 1.1 1.4 1.5 2.1 2.8 2.9 3.4 3.9 4.8 5.9 6.4 6.8 8.0 10.1 12.3 14.9 17.7 21.0 25.9 30.1 32.8 37.7 44.7 54.2 62.3
Communications
Total IT
Gross domestic product
Price
Value
Price
Value
Price
Value
Price
1.25 1.22 1.18 1.13 1.11 1.08 0.99 1.02 1.01 1.07 1.14 1.11 1.10 1.12 1.18 1.25 1.25 1.27 1.27 1.30 1.35 1.42 1.45 1.44 1.43 1.41 1.36 1.36 1.35 1.30 1.26
1.6 1.4 1.6 1.9 2.3 2.6 2.3 2.5 3.1 3.7 3.2 3.8 4.2 4.6 5.0 5.1 5.6 6.6 8.0 8.8 9.4 10.5 11.5 11.8 12.6 14.6 15.9 17.1 19.0 22.2 25.6 30.0 33.9 37.4 38.8 42.1 48.2 53.7 57.9 58.4 63.9 66.5 69.5
1.04 1.04 1.07 1.11 1.07 1.03 1.04 1.03 1.05 1.10 1.10 1.10 1.11 1.10 1.11 1.06 1.02 1.00 0.95 0.95 0.96 0.98 1.02 1.03 1.04 1.06 1.10 1.16 1.19 1.18 1.22 1.25 1.32 1.39 1.43 1.43 1.43 1.41 1.37 1.35 1.32 1.31 1.31
1.6 1.4 1.6 1.9 2.3 2.6 2.3 2.5 3.1 3.7 3.2 3.9 4.5 5.1 5.5 6.3 7.2 8.8 11.0 12.3 13.0 15.3 17.3 17.9 20.1 22.7 25.4 27.6 31.0 36.1 43.2 53.3 63.4 74.9 81.8 97.8 117.4 129.7 136.4 144.7 162.6 177.5 184.1
5.07 5.09 5.20 5.41 5.22 5.01 5.06 5.04 5.11 5.33 5.37 5.37 5.40 5.28 5.16 4.81 4.55 4.35 3.93 3.80 3.72 3.75 3.80 3.74 3.69 3.82 3.87 4.17 4.10 3.91 3.59 3.43 3.30 3.16 3.22 2.97 2.87 2.65 2.50 2.40 2.36 2.32 2.23
321.0 322.0 343.4 382.1 395.1 436.6 428.1 471.8 493.7 537.2 512.6 556.9 573.1 587.6 631.3 675.9 737.6 806.8 881.7 928.7 981.9 1,052.5 1,111.2 1,182.6 1,323.8 1,509.2 1,628.7 1,808.8 2,054.9 2,270.6 2,547.6 2,878.4 3,011.1 3,341.7 3,532.2 3,886.1 4,375.0 4,624.7 4,753.7 5,119.0 5,702.8 6,028.4 6,339.7
0.19 0.19 0.19 0.19 0.19 0.20 0.20 0.20 0.21 0.22 0.21 0.21 0.22 0.22 0.22 0.23 0.23 0.24 0.25 0.25 0.26 0.27 0.28 0.29 0.31 0.33 0.36 0.40 0.43 0.46 0.49 0.53 0.56 0.61 0.66 0.69 0.73 0.74 0.73 0.76 0.81 0.83 0.85
Ch. 10:
Accounting for Growth in the Information Age
759
Table 1 (Continued)
Computer Year
Value
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
52.5 55.2 56.3 60.4 74.9 84.8 94.1 96.6 101.9 109.9 98.5 88.3
Price 10.77 9.76 8.57 8.20 5.61 3.53 2.43 1.69 1.22 1.00 0.79 0.64
Software
Communications
Total IT
Gross domestic product
Value
Price
Value
Price
Value
Price
Value
Price
70.8 76.7 86.1 93.3 102.0 115.4 142.3 162.5 194.7 222.7 219.6 212.8
1.25 1.16 1.14 1.11 1.09 1.05 1.00 0.97 0.97 1.00 1.01 1.00
66.9 70.5 76.7 84.3 94.4 107.8 119.2 124.0 133.9 152.6 146.5 127.4
1.33 1.31 1.29 1.26 1.21 1.18 1.17 1.11 1.05 1.00 0.95 0.90
190.3 202.5 219.1 238.0 271.2 307.9 355.6 383.1 430.5 485.2 464.6 428.5
2.23 2.10 2.00 1.94 1.72 1.48 1.31 1.15 1.05 1.00 0.94 0.88
6,464.4 6,795.0 7,038.5 7,579.6 7,957.1 8,475.5 8,960.9 9,346.9 9,824.2 10,399.5 10,628.5 11,279.4
0.87 0.89 0.89 0.93 0.95 0.97 0.98 0.98 0.98 1.00 1.01 1.04
Notes: Values are in billions of current dollars. Price are normalized to one in 2000. Information technology output is final demand by type of product.
2.1.2. Capital services This section presents capital estimates for the U.S. economy for the period 1948 to 2002.21 These begin with BEA investment data; the perpetual inventory method generates estimates of capital stocks and these are aggregated, using service prices as weights. This approach, originated by Jorgenson and Zvi Griliches (1967), is based on the identification of service prices with marginal products of different types of capital. The service price estimates incorporate the cost of capital.22 The cost of capital is an annualization factor that transforms the price of an asset into the price of the corresponding capital input. This includes the nominal rate of return, the rate of depreciation, and the rate of capital loss due to declining prices. The cost of capital is an essential concept for the economics of information technology,23 due to the astonishing decline of IT prices given in Tables 1 and 2. The cost of capital is important in many areas of economics, especially in modeling producer behavior, productivity measurement, and the economics of taxation.24 Many 21 See Jorgenson and Stiroh (2000b, Appendix B), for details on the estimates of capital input. 22 Jorgenson and Kun-Young Yun (2001) present the model of capital input used in the estimates presented in
this section. BLS (1983) describes the version of this model employed in the official productivity statistics. For a recent updates, see the BLS multifactor productivity website: http://www.bls.gov/mfp/home.htm. Charles Hulten (2001) surveys the literature. 23 Jorgenson and Stiroh (1995, pp. 300–303). 24 Lawrence Lau (2000) surveys applications of the cost of capital.
760
D.W. Jorgenson Table 2 Growth rates of outputs and inputs 1989–1995
1950–2002
Prices
Quantities
Gross domestic product Information technology Computers Software Communications equipment Non-information technology investment Non-information technology consumption
2.20 −4.95 −12.68 −2.82 −1.36 2.05 2.52
2.43 12.01 17.29 13.35 7.19 1.10 2.40
Gross domestic income Information technology capital services Computer capital services Software capital services Communications equipment capital services Non-information technology capital services Labor services
2.45 −3.82 −10.46 −4.40 0.99 1.71 3.37
2.17 12.58 20.22 15.03 5.99 1.91 1.64
Prices
Quantities
Outputs 1.39 −9.58 −31.00 −1.31 −4.16 2.02 1.79 Inputs 2.10 −10.66 −26.09 −1.72 −5.56 1.72 3.42
3.59 16.12 33.36 11.82 8.44 2.01 3.35 2.88 18.33 32.34 14.27 9.83 3.01 1.50
Notes: Average annual percentage rates of growth. Information technology output is final demand by type of product. Capital services includes business capital, household consumer durables, and government capital.
of the important issues in measuring the cost of capital have been debated for decades. The first of these is incorporation of the rate of decline of asset prices into the cost of capital. The assumption of perfect foresight or rational expectations quickly emerged as the most appropriate formulation and has been used in almost all applications of the cost of capital.25 The second empirical issue is the measurement of economic depreciation. The stability of patterns of depreciation in the face of changes in tax policy and price shocks has been carefully documented. The depreciation rates presented by Jorgenson and Stiroh (2000b) summarize a large body of empirical research on the behavior of asset prices.26 A third empirical issue is the description of the tax structure for capital income. This depends on the tax laws prevailing at each point of time. The resolution of these issues has cleared the way for detailed measurements of the cost of capital for all assets that appear in the national accounts, including information technology equipment and software.27 25 See, for example, Jorgenson, Gollop, and Fraumeni (1987, pp. 40–49), and Jorgenson and Griliches (1967). 26 Jorgenson and Stiroh (2000b, Table B4, pp. 196–197) give the depreciation rates employed in this section.
Fraumeni (1997) describes depreciation rates used in the NIPA. Jorgenson (1996) surveys empirical studies of depreciation. 27 See Jorgenson and Yun (2001) for details on the U.S. tax structure for capital income. Diewert and Denis Lawrence (2000) survey measures of the price and quantity of capital input.
Ch. 10:
Accounting for Growth in the Information Age
761
The definition of capital includes all tangible assets in the U.S. economy, equipment and structures, as well as consumers’ and government durables, land, and inventories. The capital service flows from durable goods employed by households and governments enter measures of both output and input. A steadily rising proportion of these service flows are associated with investments in IT. Investments in IT by business, household, and government sectors must be included in the GDP, along with household and government IT capital services, in order to capture the full impact of IT on the U.S. economy. Table 3 gives capital stocks for the business, household and government sectors from 1948 to 2002, as well as price indexes for total domestic tangible assets and IT assets – computers, software, and communications equipment. The estimate of domestic tangible capital stock in Table 3 is $45.9 trillions in 2002, considerably greater than the estimate by BEA. The most important differences reflect the inclusion of inventories and land in Table 3. Business IT investments, as well as purchases of computers, software, and communications equipment by households and governments, have grown spectacularly in recent years, but remain relatively small. The stocks of all IT assets combined account for only 3.79 percent of domestic tangible capital stock in 2002. Table 4 presents estimates of the flow of capital services from the business, household, and government sectors and corresponding price indexes for 1948–2002. The difference between growth in capital services and capital stock is the improvement in capital quality. This represents the substitution towards assets with higher marginal products. The shift toward IT increases the quality of capital, since computers, software, and communications equipment have relatively high marginal products. Capital stock estimates fail to account for this increase in quality and substantially underestimate the impact of IT investment on growth. Table 7 shows the growth of capital quality is near twenty percent of capital input growth for the period 1948–2002. However, improvements in capital quality have increased steadily in relative importance. These improvements jumped to 46 percent of total growth in capital input during the period 1995–2002, reflecting very rapid restructuring of capital to take advantage of the sharp acceleration in the IT price decline. Capital stock has become progressively less accurate as a measure of capital input and is now seriously deficient. Figure 5 gives the IT capital service flows as a share of gross domestic income. The second panel of Table 2 summarizes the growth rates of prices and quantities of capital inputs for 1989–1995 and 1995–2002. Growth of IT capital services jumps from 12.58 percent per year in 1989–1995 to 18.33 percent in 1995–2002, while growth of nonIT capital services increases from 1.91 percent to 3.01 percent. This reverses the trend toward slower capital growth through 1995. 2.1.3. Labor services This section presents estimates of labor input for the U.S. economy from 1948 to 2002. These incorporate individual data from the Censuses of Population for 1970, 1980, and
762
D.W. Jorgenson Table 3 Information technology capital stock and domestic tangible assets
Computer Year 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989
Value
0.2 0.2 0.4 0.5 1.0 1.6 2.2 2.9 3.7 4.3 5.3 6.3 6.3 7.4 8.7 9.2 9.8 10.5 12.5 14.2 19.4 24.4 33.9 42.7 53.0 66.7 78.3 86.8 95.0 108.3 122.4
Price
2,389.62 2,389.62 1,792.22 1,194.81 836.37 716.89 597.41 414.53 333.84 283.77 258.34 232.09 176.08 142.24 134.90 110.16 101.69 85.07 73.95 50.03 41.48 32.35 28.42 25.37 21.01 17.08 14.59 12.46 10.66 9.90 9.27
Software Value
0.1 0.1 0.3 0.4 0.7 1.0 1.5 2.1 2.9 3.4 4.6 6.2 7.0 8.1 9.6 11.7 14.4 16.3 18.1 20.4 24.5 29.6 36.2 43.2 50.3 60.1 70.5 79.3 91.2 105.4 121.9
Communications
Price
Value
Price
1.16 1.16 1.14 1.10 1.06 1.04 1.02 0.94 0.97 0.96 1.01 1.09 1.06 1.05 1.07 1.12 1.19 1.19 1.21 1.21 1.24 1.29 1.35 1.39 1.38 1.37 1.36 1.31 1.31 1.30 1.25
4.6 5.7 7.0 8.6 10.0 11.5 12.9 14.3 16.4 19.4 21.1 23.1 24.9 27.1 29.9 32.0 34.5 37.8 42.1 47.9 54.4 61.7 70.0 77.3 85.1 93.8 105.8 120.6 133.0 142.2 160.3 181.8 210.4 243.3 270.5 293.2 320.5 348.1 374.1 402.8 432.8 461.6
0.93 0.93 0.95 0.99 0.96 0.92 0.93 0.92 0.94 0.98 0.98 0.98 0.96 0.94 0.94 0.92 0.91 0.89 0.88 0.89 0.91 0.93 0.96 0.98 1.01 1.02 1.07 1.14 1.18 1.16 1.19 1.22 1.28 1.36 1.40 1.41 1.42 1.42 1.40 1.39 1.37 1.36
Total IT Value 4.6 5.7 7.0 8.6 10.0 11.5 12.9 14.3 16.4 19.4 21.1 23.4 25.2 27.8 30.8 33.7 37.1 41.5 47.1 54.5 62.1 71.6 82.5 90.7 100.7 112.0 126.7 144.8 159.7 172.8 194.9 225.8 264.4 313.5 356.4 396.5 447.3 496.9 540.2 588.9 646.4 705.9
Total domestic tangible assets
Price
Value
Price
1.99 2.00 2.04 2.13 2.05 1.97 1.99 1.98 2.01 2.10 2.11 2.11 2.06 2.02 2.00 1.94 1.90 1.86 1.78 1.78 1.79 1.82 1.87 1.86 1.87 1.89 1.94 2.06 2.09 2.04 2.03 2.05 2.09 2.17 2.20 2.16 2.11 2.05 1.96 1.91 1.87 1.83
754.9 787.1 863.5 990.4 1,066.5 1,136.3 1,187.7 1,279.3 1,417.8 1,516.9 1,586.0 1,682.5 1,780.8 1,881.0 2,007.2 2,115.4 2,201.2 2,339.3 2,534.9 2,713.9 3,004.5 3,339.1 3,617.5 3,942.2 4,463.6 5,021.4 5,442.4 6,242.6 6,795.1 7,602.8 8,701.7 10,049.5 11,426.5 13,057.6 14,020.9 14,589.5 15,901.1 17,616.4 18,912.4 20,263.4 21,932.4 23,678.3
0.11 0.11 0.11 0.12 0.12 0.13 0.13 0.13 0.14 0.14 0.14 0.15 0.15 0.15 0.16 0.16 0.16 0.16 0.17 0.17 0.18 0.20 0.21 0.22 0.24 0.26 0.27 0.30 0.32 0.35 0.38 0.43 0.47 0.53 0.55 0.57 0.60 0.64 0.67 0.70 0.74 0.78
Ch. 10:
Accounting for Growth in the Information Age
763
Table 3 (Continued)
Computer Year
Value
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
123.6 125.7 129.7 138.9 155.5 178.3 192.5 212.5 227.4 252.1 288.2 279.5 281.8
Software
Price 8.30 7.38 6.20 5.22 4.59 3.80 2.85 2.15 1.55 1.18 1.00 0.80 0.67
Communications
Total IT
Total domestic tangible assets
Value
Price
Value
Price
Value
Price
Value
Price
140.6 163.2 175.0 199.2 218.2 242.7 269.7 312.3 360.6 433.6 515.5 563.7 583.9
1.22 1.22 1.12 1.11 1.08 1.07 1.04 1.00 0.97 0.97 1.00 1.01 1.00
487.4 508.0 528.8 550.6 578.0 605.5 637.5 678.7 704.3 741.3 805.2 844.3 874.0
1.35 1.34 1.32 1.30 1.28 1.24 1.20 1.18 1.11 1.05 1.00 0.95 0.91
751.6 796.9 833.4 888.7 951.7 1,026.5 1,099.7 1,203.5 1,292.2 1,427.1 1,608.9 1,687.5 1,739.7
1.77 1.73 1.64 1.58 1.52 1.44 1.34 1.25 1.13 1.05 1.00 0.94 0.89
24,399.0 24,896.3 25,218.3 25,732.9 26,404.2 28,003.7 29,246.9 31,146.2 33,888.6 36,307.5 39,597.1 42,566.9 45,892.1
0.79 0.79 0.79 0.79 0.79 0.82 0.83 0.86 0.91 0.95 1.00 1.05 1.11
Notes: Values are in billions of current dollars. Price are normalized to one in 2000. Domestic tangible assets include fixed assets and consumer durable goods, land, and inventories.
1990, as well as the annual Current Population Surveys. Constant quality indexes for the price and quantity of labor input account for the heterogeneity of the workforce across sex, employment class, age, and education levels. This follows the approach of Jorgenson, Gollop, and Fraumeni (1987).28 The distinction between labor input and labor hours is analogous to the distinction between capital services and capital stock. The growth in labor quality is the difference between the growth in labor input and hours worked. Labor quality reflects the substitution of workers with high marginal products for those with low marginal products. Table 5 presents estimates of labor input, hours worked, and labor quality. The value of labor expenditures in Table 5 is $6.6 trillions in 2002, 58.3 percent of the value of output. This share accurately reflects the concept of gross domestic income, including imputations for the value of capital services in household and government sectors. As shown in Table 7, the growth rate of labor input decelerated to 1.50 percent for 1995–2002 from 1.64 percent for 1989–1995. Growth in hours worked rose from 1.02 percent for 1989–1995 to 1.16 percent for 1995–2002 as labor force participation increased and unemployment rates declined. The growth of labor quality has declined considerably since 1995, dropping from 0.61 percent for 1989–1995 to 0.33 percent for 1995–2002. This slowdown captures wellknown demographic trends in the composition of the work force, as well as exhaustion 28 See Jorgenson and Stiroh (2000b, Appendix C), for details on the estimates of labor input. Gollop (2000)
discusses the measurement of labor quality.
764
D.W. Jorgenson Table 4 Information technology capital services and gross domestic income
Computer Year 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991
Value
0.2 0.2 0.3 0.5 0.7 0.8 1.2 2.2 2.4 2.6 2.7 3.6 5.2 4.9 4.4 6.6 5.9 6.6 7.0 11.8 11.6 16.6 17.6 19.7 26.6 36.5 40.0 43.6 54.0 53.4 58.6 65.9 65.8
Price
1,842.99 1,801.00 2,651.34 2,221.36 1,301.75 749.35 680.70 675.44 424.81 329.85 253.51 246.32 274.21 182.44 124.14 145.93 107.43 97.74 78.00 84.61 50.14 44.40 29.68 22.39 20.57 18.42 14.01 11.46 11.00 8.69 7.83 7.57 6.65
Software Value
0.1 0.1 0.1 0.2 0.3 0.4 0.6 0.9 1.1 1.5 1.7 2.3 3.5 3.7 4.2 4.9 6.2 7.0 7.7 9.0 10.4 12.2 13.7 15.2 18.0 22.4 26.7 31.0 36.5 44.6 54.3 60.0 62.7
Communications
Total IT
Gross domestic income
Price
Value
Price
Value
Price
Value
Price
1.54 1.51 1.53 1.59 1.40 1.32 1.36 1.43 1.14 1.27 1.11 1.17 1.52 1.37 1.32 1.34 1.46 1.44 1.43 1.50 1.50 1.51 1.45 1.39 1.40 1.46 1.46 1.44 1.46 1.54 1.57 1.45 1.29
1.7 1.4 1.7 2.0 2.6 3.1 2.5 3.5 4.0 3.5 3.9 4.9 5.1 5.3 6.3 6.2 6.8 8.7 9.2 9.4 9.8 10.9 12.7 14.3 16.0 21.7 19.5 22.3 23.9 39.4 33.6 44.9 40.0 38.6 41.4 46.4 53.9 60.3 67.3 78.8 97.2 98.8 102.4 97.0
1.21 0.87 0.86 0.91 0.97 0.98 0.70 0.87 0.89 0.69 0.70 0.81 0.76 0.72 0.77 0.68 0.69 0.80 0.75 0.68 0.64 0.64 0.68 0.70 0.73 0.92 0.76 0.82 0.82 1.26 0.98 1.19 0.96 0.84 0.83 0.87 0.93 0.96 0.98 1.06 1.20 1.14 1.10 0.99
1.7 1.4 1.7 2.0 2.6 3.1 2.5 3.5 4.0 3.5 3.9 5.2 5.3 5.7 7.0 7.1 7.9 10.5 12.4 12.8 14.0 15.3 18.5 23.0 24.5 30.2 30.9 34.4 37.5 54.2 54.4 66.8 68.8 69.9 76.3 91.1 112.8 127.0 141.9 169.3 195.2 211.6 228.3 225.6
8.44 6.12 6.04 6.36 6.78 6.85 4.88 6.08 6.24 4.84 4.88 5.66 5.33 5.15 5.41 4.63 4.40 4.96 4.73 3.96 3.66 3.44 3.58 3.87 3.58 3.91 3.56 3.56 3.51 4.54 3.92 4.00 3.41 2.84 2.61 2.61 2.64 2.45 2.32 2.38 2.39 2.27 2.17 1.93
321.0 322.0 343.4 382.1 395.1 436.6 428.1 471.8 493.7 537.2 512.6 556.9 573.1 587.6 631.3 675.9 737.6 806.8 881.7 928.7 981.9 1,052.5 1,111.2 1,182.6 1,323.8 1,509.2 1,628.7 1,808.8 2,054.9 2,270.6 2,547.6 2,878.4 3,011.1 3,341.7 3,532.2 3,886.1 4,375.0 4,624.7 4,753.7 5,119.0 5,702.8 6,028.4 6,339.7 6,464.4
0.14 0.13 0.14 0.14 0.14 0.15 0.15 0.16 0.16 0.17 0.16 0.17 0.17 0.17 0.18 0.19 0.20 0.21 0.22 0.22 0.22 0.23 0.24 0.26 0.28 0.30 0.32 0.36 0.39 0.42 0.45 0.49 0.51 0.55 0.58 0.64 0.68 0.69 0.70 0.72 0.77 0.79 0.81 0.82
Ch. 10:
Accounting for Growth in the Information Age
765
Table 4 (Continued)
Computer Year
Value
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
73.2 79.4 84.0 105.2 133.1 144.0 162.4 166.5 156.9 175.6 162.9
Software
Price 6.22 5.38 4.46 4.18 3.73 2.77 2.12 1.48 1.00 0.88 0.67
Communications
Total IT
Gross domestic income
Value
Price
Value
Price
Value
Price
Value
Price
82.4 80.0 97.3 102.7 116.3 134.4 152.5 165.7 193.8 219.0 247.2
1.45 1.21 1.29 1.21 1.20 1.17 1.10 1.00 1.00 1.01 1.07
105.4 118.1 132.6 150.2 144.2 147.6 184.4 188.1 201.4 199.5 202.4
1.02 1.09 1.14 1.20 1.07 1.01 1.15 1.06 1.00 0.88 0.82
261.0 277.5 313.8 358.0 393.5 425.9 499.3 520.2 552.1 594.1 612.5
1.99 1.85 1.83 1.80 1.66 1.46 1.37 1.15 1.00 0.92 0.86
6,795.0 7,038.5 7,579.6 7,957.1 8,475.5 8,960.9 9,346.9 9,824.2 10,399.5 10,628.5 11,279.4
0.85 0.86 0.90 0.91 0.94 0.96 0.96 0.98 1.00 1.00 1.06
Notes: Values are in billions of current dollars. Prices are normalized to one in 2000.
of the pool of available workers. Growth in hours worked does not capture these changes in labor quality growth and is a seriously misleading measure of labor input. 2.2. The American growth resurgence The American economy has undergone a remarkable resurgence since the mid-1990s with accelerating growth in output, labor productivity, and total factor productivity. The purpose of this section is to quantify the sources of growth for 1948–2002 and various sub-periods. An important objective is to account for the sharp acceleration in the growth rate since 1995 and, in particular, to document the role of information technology. The appropriate framework for analyzing the impact of information technology is the production possibility frontier, giving outputs of IT investment goods as well as inputs of IT capital services. An important advantage of this framework is that prices of IT outputs and inputs are linked through the price of IT capital services. This framework successfully captures the substitutions among outputs and inputs in response to the rapid deployment of IT. It also encompasses costs of adjustment, while allowing financial markets to be modeled independently. As a consequence of the swift advance of information technology, a number of the most familiar concepts in growth economics have been superseded. The aggregate production function heads this list. Capital stock as a measure of capital input is no longer adequate to capture the rising importance of IT. A stock measure completely obscures the restructuring of capital input that is such an important wellspring of the growth resurgence. Finally, hours worked must be replaced as a measure of labor input.
766
D.W. Jorgenson Table 5 Labor services Labor services
Weekly
Hourly
Hours
Year
Price
Quantity
Value
Quality
Employment
hours
compensation
worked
1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991
0.06 0.07 0.08 0.08 0.09 0.09 0.09 0.09 0.10 0.11 0.12 0.11 0.12 0.12 0.13 0.13 0.13 0.14 0.15 0.16 0.17 0.18 0.20 0.22 0.23 0.25 0.27 0.29 0.32 0.34 0.37 0.40 0.44 0.47 0.50 0.54 0.56 0.58 0.64 0.64 0.65 0.68 0.71 0.75
2,324.8 2,262.8 2,350.6 2,531.5 2,598.2 2,653.0 2,588.7 2,675.7 2,738.0 2,740.9 2,671.8 2,762.8 2,806.6 2,843.4 2,944.4 2,982.3 3,055.7 3,149.7 3,278.8 3,327.2 3,405.4 3,491.1 3,439.2 3,439.5 3,528.8 3,672.4 3,660.9 3,606.4 3,708.0 3,829.8 3,994.9 4,122.6 4,105.6 4,147.7 4,110.2 4,172.3 4,417.4 4,531.7 4,567.5 4,736.5 4,888.8 5,051.3 5,137.6 5,086.7
150.1 165.5 181.3 210.7 222.5 238.5 240.7 252.7 272.4 293.0 307.4 316.9 341.7 352.1 374.1 382.7 412.0 448.1 494.8 518.9 582.6 641.4 683.1 740.7 813.3 903.9 979.2 1,055.2 1,182.6 1,321.1 1,496.8 1,660.4 1,809.7 1,934.6 2,056.5 2,234.7 2,458.3 2,646.2 2,904.1 3,017.3 3,173.3 3,452.4 3,673.2 3,806.3
0.73 0.73 0.75 0.76 0.77 0.79 0.79 0.80 0.80 0.81 0.82 0.82 0.83 0.84 0.85 0.86 0.86 0.86 0.87 0.87 0.88 0.88 0.88 0.89 0.89 0.89 0.89 0.90 0.90 0.90 0.90 0.90 0.90 0.91 0.92 0.92 0.93 0.93 0.93 0.94 0.94 0.95 0.96 0.96
61,536 60,437 62,424 66,169 67,407 68,471 66,843 68,367 69,968 70,262 68,578 70,149 71,128 71,183 72,673 73,413 74,990 77,239 80,802 82,645 84,733 87,071 86,867 86,715 88,838 92,542 94,121 92,575 94,922 98,202 102,931 106,463 107,061 108,050 106,749 107,810 112,604 115,201 117,158 120,456 123,916 126,743 128,290 127,022
40.6 40.2 39.8 39.7 39.2 38.8 38.4 38.7 38.4 37.9 37.6 37.8 37.6 37.4 37.4 37.3 37.2 37.2 36.8 36.3 36.0 35.9 35.3 35.2 35.3 35.2 34.5 34.2 34.2 34.1 34.0 33.9 33.4 33.3 33.1 33.2 33.3 33.3 33.0 33.1 33.0 33.2 33.0 32.7
1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.8 1.9 2.1 2.3 2.3 2.5 2.5 2.6 2.7 2.8 3.0 3.2 3.3 3.7 3.9 4.3 4.7 5.0 5.3 5.8 6.4 7.0 7.6 8.2 8.9 9.7 10.4 11.2 12.0 12.6 13.3 14.4 14.6 14.9 15.8 16.7 17.6
129,846 126,384 129,201 136,433 137,525 138,134 133,612 137,594 139,758 138,543 134,068 137,800 139,150 138,493 141,258 142,414 144,920 149,378 154,795 156,016 158,604 162,414 159,644 158,943 162,890 169,329 168,800 164,460 168,722 174,265 181,976 187,589 186,202 186,887 183,599 186,175 195,221 199,424 200,998 207,119 212,882 218,811 220,475 216,281
Ch. 10:
Accounting for Growth in the Information Age
767
Table 5 (Continued) Weekly
Hourly
Hours
Year
Price
Quantity
Labor services Value
Quality
Employment
hours
compensation
worked
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
0.80 0.82 0.83 0.84 0.86 0.89 0.92 0.96 1.00 1.05 1.06
5,105.9 5,267.6 5,418.2 5,573.2 5,683.6 5,843.3 6,020.8 6,152.1 6,268.5 6,250.6 6,188.7
4,087.4 4,323.8 4,472.4 4,661.5 4,878.5 5,186.5 5,519.5 5,908.2 6,268.5 6,537.4 6,576.2
0.97 0.97 0.98 0.98 0.99 0.99 0.99 1.00 1.00 1.00 1.01
127,100 129,556 132,459 135,297 137,571 140,432 143,557 146,468 149,364 149,020 147,721
32.8 32.9 33.0 33.1 33.0 33.2 33.3 33.3 33.1 32.9 32.9
18.8 19.5 19.7 20.0 20.7 21.4 22.2 23.3 24.4 25.6 26.1
216,873 221,699 227,345 232,675 235,859 242,242 248,610 253,276 257,048 255,054 252,399
Notes: Value is in billions of current dollars. Quantity is in billions of 2000 dollars. Price and quality are normalized to one in 2000. Employment is in thousands of workers. Weekly hours is hours per worker, divided by 52. Hourly compensation is in current dollars. Hours worked are in millions of hours.
2.2.1. Production possibility frontier The production possibility frontier describes efficient combinations of outputs and inputs for the economy as a whole. Aggregate output Y consists of outputs of investment goods and consumption goods. These outputs are produced from aggregate input X, consisting of capital services and labor services. Productivity is a “Hicks-neutral” augmentation of aggregate input. The production possibility frontier takes the form: Y (Yn , Yc , Ys , Ym ) = A · f (Kn , Kc , Ks , Km , L)
(1)
where the outputs include non-IT output goods Yn and output of computers Yc , software Ys , and communications equipment Ym . Inputs include non-IT capital services Kn and the services of computers Kc , software Ks , and telecommunications equipment Km , as well as labor input L.29 Total factor productivity is denoted by A. The most important advantage of the production possibility frontier is the explicit role that it provides for constant quality prices of IT products. These are used as deflators for nominal expenditures on IT investments to obtain the quantities of IT outputs. Investments in IT are cumulated into stocks of IT capital. The flow of IT capital services is an aggregate of these stocks with service prices as weights. Similarly, constant quality prices of IT capital services are used in deflating the nominal values of consumption of these services. 29 Services of durable goods to governments and households are included in both inputs and outputs.
768
D.W. Jorgenson
Another important advantage of the production possibility frontier is the incorporation of costs of adjustment. For example, an increase in the output of IT investment goods requires foregoing part of the output of consumption goods and non-IT investment goods, so that adjusting the rate of investment in IT is costly. However, costs of adjustment are external to the producing unit and are fully reflected in IT prices. These prices incorporate forward-looking expectations of the future prices of IT capital services. The aggregate production function employed, for example, by Simon Kuznets (1971) and Robert Solow (1957, 1960, 1970) and, more recently, by Jeremy Greenwood, Zvi Hercowitz and Per Krusell (1997, 2000), Hercowitz (1998), and Arnold Harberger (1998) is a competing methodology. The production function gives a single output as a function of capital and labor inputs. There is no role for separate prices of investment and consumption goods and, hence, no place for constant quality IT price indexes for outputs of IT investment goods. Another limitation of the aggregate production function is that it fails to incorporate costs of adjustment. Robert Lucas (1967) presented a production model with internal costs of adjustment. Fumio Hayashi (2000) shows how to identify these adjustment costs from James Tobin’s (1969) Q-ratio, the ratio of the stock market value of the producing unit to the market value of the unit’s assets. Implementation of this approach requires simultaneous modeling of production and asset valuation. If costs of adjustment are external, as in the production possibility frontier, asset valuation can be modeled separately from production.30 2.2.2. Sources of growth Under the assumption that product and factor markets are competitive producer equilibrium implies that the share-weighted growth of outputs is the sum of the share-weighted growth of inputs and growth in total factor productivity: wY,n ln Yn + wY,c Yc + wY,s Ys + wY,m Ym = v K,n ln Kn + v K,c ln Kc + v K,s ln Ks + v K,m ln Km + v L ln L + ln A,
(2)
where w and v denote average value shares of the subscripted variable. The shares of outputs and inputs add to one under the additional assumption of constant returns, wY,n + wY,c + wY,s + wY,m = v K,n + v K,c + v K,s + v K,m + v L = 1. The growth rate of output is a weighted average of growth rates of investment and consumption goods outputs. The contribution of each output is its weighted growth rate. Similarly, the growth rate of input is a weighted average of growth rates of capital and labor services and the contribution of each input is its weighted growth rate. The contribution of total factor productivity, the growth rate of the augmentation. 30 See, for example, John Campbell and Robert Shiller (1998).
Ch. 10:
Accounting for Growth in the Information Age
769
Table 6 Sources of gross domestic product growth 1948–02 1948–73 1973–89 1989–95 1995–02
Gross domestic product Contribution of information technology Computers Software Communications equipment Contribution of non-information technology Contribution of non-information technology investment Contribution of non-information technology consumption
3.46 0.28 0.13 0.07 0.08 3.18 0.69 2.49
3.99 0.11 0.03 0.02 0.07 3.88 1.05 2.82
Gross domestic income Contribution of information technology capital services Computers Software Communications equipment Contribution of non-information technology capital services Contribution of labor services Total factor productivity
2.79 0.36 0.17 0.08 0.11 1.39 1.05 0.67
2.99 0.15 0.04 0.02 0.09 1.79 1.04 1.00
Outputs 2.97 0.35 0.18 0.08 0.09 2.62 0.44 2.18 Inputs 2.68 0.38 0.20 0.07 0.11 1.15 1.15 0.29
2.43 0.37 0.15 0.15 0.08 2.05 0.21 1.85
3.59 0.64 0.34 0.19 0.11 2.95 0.41 2.54
2.17 0.49 0.22 0.16 0.10 0.71 0.98 0.26
2.88 0.93 0.52 0.23 0.18 1.07 0.88 0.71
Notes: Average annual percentage rates of growth. The contribution of an output or input is the rate of growth, multiplied by the average value share.
Table 6 presents results of a growth accounting decomposition for the period 1948– 2002 and various sub-periods, following Jorgenson and Stiroh (1999, 2000b). Economic growth is broken down by output and input categories, quantifying the contribution of information technology to outputs, as well as capital inputs. These estimates identify computer hardware, software, and communications equipment as distinct types of information technology. The results can also be presented in terms of average labor productivity (ALP), defined as y = Y/H , the ratio of output Y to hours worked H , and k = K/H is the ratio of capital services K to hours worked: ln y = v K ln k + v L ( ln L − ln H ) + ln A.
(3)
This equation allocates ALP growth among three sources. The first is capital deepening, the growth in capital input per hour worked, and reflects the capital–labor substitution. The second is improvement in labor quality and captures the rising proportion of hours by workers with higher marginal products. The third is total factor productivity growth, which contributes point-for-point to ALP growth. Table 7 shows these estimates.
770
D.W. Jorgenson Table 7 Sources of average labor productivity growth 1948–02
Gross domestic product Hours worked Average labor productivity Contribution of capital deepening Information technology Non-information technology Contribution of labor quality Total factor productivity Information technology Non-information technology
3.46 1.23 2.23 1.23 0.33 0.90 0.33 0.67 0.17 0.50
Labor input Labor quality Capital input Capital stock Capital quality
1.81 0.58 4.13 3.29 0.84
1948–73
1973–89
1989–95
3.99 2.97 2.43 1.06 1.60 1.02 2.93 1.36 1.40 1.49 0.85 0.78 0.14 0.34 0.44 1.35 0.51 0.34 0.43 0.23 0.36 1.00 0.29 0.26 0.05 0.20 0.23 0.95 0.09 0.03 Addendum – Growth rates 1.83 1.99 1.64 0.77 0.39 0.61 4.49 3.67 2.92 4.13 2.77 1.93 0.36 0.90 0.99
1995–02 3.59 1.16 2.43 1.52 0.88 0.64 0.20 0.71 0.47 0.24 1.50 0.33 4.92 2.66 2.27
Notes: Average annual percentage rates of growth. Contributions are defined in Equation (3).
2.2.3. Contributions of IT output Figure 5 depicts the rapid increase in the importance of IT services, reflecting the accelerating pace of IT price declines. In 1995–2002 the capital service price for computers fell 26.09 percent per year, compared to an increase of 32.34 percent in capital input from computers. While the value of computer services grew, the current dollar value was only 1.44 percent of gross domestic income in 2002. The rapid accumulation of software appears to have different sources. The price of software services has fallen only 1.72 percent per year for 1995–2002. Nonetheless, firms have been accumulating software very rapidly, with real capital services growing 14.27 percent per year. A possible explanation is that firms respond to computer price declines by investing in complementary inputs like software. However, a more plausible explanation is that the price indexes used to deflate software investment fail to hold quality constant. This leads to an overstatement of inflation and an understatement of growth. Although the price decline for communications equipment during the period 1995– 2002 is greater than that of software, investment in this equipment is more in line with prices. However, prices of communications equipment also fail to hold quality constant. The technology of switching equipment, for example, is similar to that of computers; investment in this category is deflated by a constant-quality price index developed by BEA. Conventional price deflators are employed for transmission gear, such as fiber-
Ch. 10:
Accounting for Growth in the Information Age
771
Figure 5. Input shares of information technology by type, 1948–2002. Note: Share of current dollar gross domestic income.
optic cables. This leads to an underestimate of the growth rates of investment, capital stock, capital services, and the GDP, as well as an overestimate of the rate of inflation. Figures 6 and 7 highlight the rising contributions IT outputs to U.S. economic growth. Figure 6 shows the breakdown between IT and non-IT outputs for sub-periods from 1948 to 2002, while Figure 7 decomposes the contribution of IT into its components. Although the importance of IT has steadily increased, Figure 6 shows that the recent investment and consumption surge nearly doubled the output contribution of IT. Figure 7 shows that computer investment is the largest single IT contributor after 1995, but that investments in software and communications equipment are becoming increasingly important. Table 2 reports IT input prices and Figures 8 and 9 present a similar decomposition of IT inputs into production. The contribution of these inputs is rising even more dramatically. Figure 8 shows that the contribution of IT now accounts for more than 45.0 percent of the total contribution of capital input. Figure 9 reveals that computer hardware is the largest component of IT, reflecting the growing share and accelerating growth rate of computer investment in the late 1990s. Private business investment predominates in the output of IT, as shown by Jorgenson and Stiroh (2000b) and Oliner and Sichel (2000).31 Household purchases of IT equipment are next in importance. Government purchases of IT equipment and net exports of
31 Bosworth and Triplett (2000) and Martin Baily (2002) compare the results of Jorgenson and Stiroh with
those of Oliner and Sichel, who incorporate data from the BLS measures of multifactor productivity.
772
D.W. Jorgenson
Figure 6. Output contribution of information technology. Note: Output contributions are the average annual growth rates, weighted by the output shares.
Figure 7. Output contribution of information technology by type. Note: Output contributions are the average annual growth rates, weighted by the output shares.
IT products must be included in order to provide a complete picture. Firms, consumers, governments, and purchasers of U.S. exports are responding to relative price changes, increasing the contributions of computers, software, and communications equipment.
Ch. 10:
Accounting for Growth in the Information Age
773
Figure 8. Capital input contribution of information technology. Note: Input contributions are the average annual growth rates, weighted by the income shares.
Figure 9. Capital input contribution of information technology by type. Note: Input contributions are the average annual growth rates, weighted by the income shares.
Table 2 shows that the price of computer investment fell by 31.00 percent per year, the price of software fell by 1.31 percent and the price of communications equipment dropped by 4.16 percent during the period 1995–2002, while non-IT investment goods
774
D.W. Jorgenson
prices rose 2.02 percent. In response to these price changes, firms, households, and governments have accumulated computers, software, and communications equipment much more rapidly than other forms of capital. 2.2.4. Total factor productivity The price or “dual” approach to productivity measurement employed by Triplett (1996) makes it possible to identify the role of IT production as a source of total factor productivity growth at the industry level.32 The rate of total factor productivity growth is measured as the decline in the price of output, plus a weighted average of the growth rates of input prices with value shares of the inputs as weights. For the computer industry this expression is dominated by two terms: the decline in the price of computers and the contribution of the price of semiconductors. For the semiconductor industry the expression is dominated by the decline in the price of semiconductors.33 Jorgenson, Gollop and Fraumeni (1987) have employed Evsey Domar’s (1961) model to trace aggregate productivity growth to its sources at the level of individual industries.34 More recently, Harberger (1998), William Gullickson and Michael Harper (1999), and Jorgenson and Stiroh (2000a, 2000b) have used the model for similar purposes. Total factor productivity growth for each industry is weighted by the ratio of the gross output of the industry to GDP to estimate the industry contribution to aggregate productivity growth. If semiconductor output were only used to produce computers, then its contribution to computer industry productivity growth, weighted by computer industry output, would precisely offset its independent contribution to the growth of aggregate productivity. This is the ratio of the value of semiconductor output to GDP, multiplied by the rate of semiconductor price decline. In fact, semiconductors are used to produce telecommunications equipment and many other products. However, the value of semiconductor output is dominated by inputs into IT production. The Domar aggregation formula can be approximated by expressing the declines in prices of computers, communications equipment, and software relative to the price of gross domestic income, an aggregate of the prices of capital and labor services. The rates of relative IT price decline are weighted by ratios of the outputs of IT products to the GDP. Table 8 reports details of this decomposition of total factor productivity; the IT and non-IT contributions are presented in Figure 10. Production of IT products contributes 0.47 percentage points to total factor productivity growth for 1995–2002, compared to 0.23 percentage points for 1989–1995. This reflects the accelerating decline in relative price changes resulting from shortening the product cycle for semiconductors. 32 The dual approach is presented by Jorgenson, Gollop, and Fraumeni (1987, pp. 53–63). 33 Models of the relationships between computer and semiconductor industries presented by Dulberger
(1993), Triplett (1996), and Oliner and Sichel (2000) are special cases of the Domar (1961) aggregation scheme. 34 See Jorgenson, Gollop and Fraumeni (1987, pp. 63–66, 301–322).
Ch. 10:
Accounting for Growth in the Information Age
775
Table 8 Sources of total factor productivity growth 1948–02 Total factor productivity growth
0.67
Information technology Computers Software Communications equipment Non-information technology
0.17 0.10 0.02 0.04 0.50
Information technology Computers Software Communications equipment Non-information technology
−7.05 −22.50 −4.87 −4.07 −0.51
Information technology Computers Software Communications equipment Non-information technology
1.94 0.46 0.53 0.95 98.06
1948–73
1973–89
1989–95
1.00 0.29 0.26 Contributions to TFP growth 0.05 0.20 0.23 0.02 0.13 0.13 0.00 0.03 0.06 0.03 0.05 0.04 0.95 0.09 0.03 Relative price changes −4.3 −9.1 −7.4 −22.0 −21.5 −15.1 −5.1 −5.1 −5.3 −3.1 −4.6 −3.8 −1.0 −0.1 0.0 Average nominal shares 0.91 2.20 3.04 0.10 0.64 0.83 0.07 0.49 1.13 0.74 1.07 1.09 99.09 97.80 96.96
1995–02 0.71 0.47 0.33 0.06 0.08 0.24 −11.7 −33.1 −3.4 −6.3 −0.3 4.10 1.00 1.78 1.33 95.90
Notes: Average annual rates of growth. Prices are relative to the price of gross domestic income. Contributions are relative price changes, weighted by average nominal output shares.
Figure 10. Contribution of information technology to total factor productivity growth. Note: Contributions are average annual relative price changes,weighted by average normal output shares from Table 8.
776
D.W. Jorgenson
Figure 11. Sources of gross domestic product growth.
2.2.5. Output growth This section presents the sources of GDP growth for the entire period 1948 to 2002 as described in Tables 6 and 7. Output grew 3.46 percent per year, as capital services contributed 1.75 percentage points, labor services 1.05 percentage points, and total factor productivity growth only 0.67 percentage points. Input growth is the source of nearly 80.6 percent of U.S. growth over the past half century, while productivity has accounted for 19.4 percent. Figure 11 shows the relatively modest contributions of productivity in all sub-periods. More than four-fifths of the contribution of capital reflects the accumulation of capital stock, while improvement in the quality of capital accounts for about one-fifth. Similarly, increased labor hours account for 68 percent of labor’s contribution; the remainder is due to improvements in labor quality. Substitutions among capital and labor inputs in response to price changes are essential components of the sources of economic growth. A look at the U.S. economy before and after 1973 reveals familiar features of the historical record. After strong output and productivity growth in the 1950s, 1960s and early 1970s, the U.S. economy slowed markedly through 1989, with output growth falling from 3.99 percent to 2.97 percent and total factor productivity growth declining from 1.00 percent to 0.29 percent for 1973 to 1989. The contribution of capital input also slowed from 1.94 percent for 1948–1973 to 1.53 percent for 1973–1989. This contributed to sluggish ALP growth – 2.93 percent for 1948–1973 compared to 1.36 percent for 1973–1989. Relative to the period 1989–1995, output growth increased by 1.16 percent during 1995–2002. The contribution of IT production jumped by 0.27 percent, relative to 1989–
Ch. 10:
Accounting for Growth in the Information Age
777
Figure 12. Sources of average labor productivity growth.
1995, but still accounted for only 17.8 percent of the growth of output. Although the contribution of IT has increased steadily throughout the period 1948–2002, there has been a sharp response to the acceleration in the IT price decline in 1995. Nonetheless, more than eighty percent of the output growth can be attributed to non-IT products. Between 1989–1995 and 1995–2002 the contribution of capital input jumped by 0.80 percentage points, the contribution of labor input declined by 0.10 percent, and total factor productivity accelerated by 0.45 percent. Growth in ALP rose 1.03 percent as more rapid capital deepening and growth in total factor productivity offset slower improvement in labor quality. Growth in hours worked slowed as labor markets tightened considerably, even as labor force participation rates increased.35 The contribution of capital input reflects the investment boom of the late 1990s as businesses, households, and governments poured resources into plant and equipment, especially computers, software, and communications equipment. The contribution of capital, predominantly IT, is considerably more important than the contribution of labor. The contribution of IT capital services has grown steadily throughout the period 1948– 2002, but Figure 9 reflects the impact of the accelerating decline in IT prices. After maintaining an average rate of 0.29 percent for the period 1973–1989, total factor productivity growth declined to 0.26 percent for 1989–1995 and then increased to 0.71 percent per year for 1995–2002. This is an increasing source of growth in output and ALP for the U.S. economy (Figures 11 and 12). Total factor productivity growth for 1995–2002 is still below the rate of 1948–1973 and the U.S. economy is recuperating from the anemic productivity growth of the past two decades. Slightly more than half of
35 Lawrence Katz and Alan Krueger (1999) analyze the recent performance of the U.S. labor market.
778
D.W. Jorgenson
the acceleration in productivity from 1989–1995 to 1995–2002 can be attributed to IT production, which is far greater than the 5.01 percent share of IT in the GDP in 2002. 2.2.6. Average labor productivity Output growth is the sum of growth in hours and average labor productivity. Table 7 shows the breakdown between growth in hours and ALP for the same periods as in Table 6. For the period 1948–2002, ALP growth predominated in output growth, increasing 2.23 percent per year, while hours worked increased 1.23 percent per year. As shown above, ALP growth depends on capital deepening, a labor quality effect, and overall productivity growth. Figure 12 reveals the well-known productivity slowdown of the 1970s and 1980s, emphasizing the sharp acceleration in labor productivity growth in the late 1990s. The slowdown through 1989 reflects reduced capital deepening, declining labor quality growth, and decelerating growth in total factor productivity. The growth of ALP recovered slightly during the early 1990s with a slump in capital deepening more than offset by a revival in labor quality growth. A slowdown in hours combined with middling ALP growth during 1989–1995 to produce a further slide in the growth of output. In previous cyclical recoveries during the postwar period, output growth accelerated during the recovery, powered by more rapid growth of hours and ALP. Accelerating output growth during 1995–2002 reflects modest growth in labor hours and a sharp increase in ALP growth.36 Comparing 1989–1995 to 1995–2002, the rate of output growth jumped by 1.16 percent – due to an increase in hours worked of 0.14 percent and an upward bound in ALP growth of 1.03 percent. Figure 12 shows the acceleration in ALP growth is due to capital deepening as well as faster total factor productivity growth. Capital deepening contributed 0.74 percentage points to the change, counterbalancing a negative contribution of labor quality of 0.16 percent. The acceleration in total factor productivity growth added 0.45 percentage points. 2.2.7. Research opportunities The use of computers, software, and communications equipment must be carefully distinguished from the production of IT.37 Massive increases in computing power, like those experienced by the U.S. economy, have two effects on growth. First, as IT producers become more efficient, more IT equipment and software is produced from the same inputs. This raises total factor productivity in IT-producing industries and contributes to productivity growth for the economy as a whole. Labor productivity also grows at both industry and aggregate levels. Second, investment in information technology leads to growth of productive capacity in IT-using industries. Since labor is working with more and better equipment, 36 Stiroh (2002) shows that ALP growth is concentrated in IT-producing and IT-using industries. 37 Economics and Statistics Administration (2000, Table 3.1, p. 23), lists IT-producing industries.
Ch. 10:
Accounting for Growth in the Information Age
779
this increases ALP through capital deepening. If the contributions to aggregate output are captured by capital deepening, aggregate total factor productivity growth is unaffected.38 Increasing deployment of IT affects productivity growth only if there are spillovers from IT-producing industries to IT-using industries. Jorgenson, Ho and Stiroh (2004) trace the increase in aggregate productivity growth to its sources in individual industries. Jorgenson and Stiroh (2000a, 2000b) present the appropriate methodology and preliminary results. Stiroh (2002) shows that aggregate ALP growth can be attributed to productivity growth in IT-producing and IT-using industries. 2.3. Demise of traditional growth accounting 2.3.1. Introduction The early 1970s marked the emergence of a rare professional consensus on economic growth, articulated in two strikingly dissimilar books. Kuznets summarized his decades of empirical research in Economic Growth of Nations (1971).39 Solow’s book Economic Growth (1970), modestly subtitled “An Exposition”, contained his 1969 Radcliffe Lectures at the University of Warwick. In these lectures Solow also summarized decades of theoretical research, initiated by the work of Roy Harrod (1939) and Domar (1946).40 Let me first consider the indubitable strengths of the perspective on growth that emerged victorious over its many competitors in the early 1970s. Solow’s neo-classical theory of economic growth, especially his analysis of steady states with constant rates of growth, provided conceptual clarity and sophistication. Kuznets generated persuasive empirical support by quantifying the long sweep of historical experience of the United States and thirteen other developed economies. He combined this with quantitative comparisons among a developed and developing economies during the postwar period. With the benefit of hindsight the most obvious deficiency of the traditional framework of Kuznets and Solow was the lack of a clear connection between the theoretical and the empirical components. This lacuna can be seen most starkly in the total absence of cross references between the key works of these two great economists. Yet they were
38 Baily and Gordon (1988). 39 The enormous impact of this research was recognized in the same year by the Royal Swedish Academy of
Sciences in awarding the third Bank of Sweden Prize in Economic Sciences in Memory of Alfred Nobel to Kuznets “for his empirically founded interpretation of economic growth which has led to new and deepened insight into the economic and social structure and process of development.” See Assar Lindbeck (1992, p. 79). 40 Solow’s seminal role in this research, beginning with his brilliant and pathbreaking essay of 1956, “A Contribution to the Theory of Economic Growth”, was recognized, simply and elegantly, by the Royal Swedish Academy of Sciences in awarding Solow the Nobel Prize in Economics in 1987 “for his contributions to the theory of economic growth”. See Karl-Goran Maler (1992, p. 191). Solow (1999) presents an updated version of his exposition of growth theory.
780
D.W. Jorgenson
working on the same topic, within the same framework, at virtually the same time, and in the very same geographical location – Cambridge, Massachusetts! Searching for analogies to describe this remarkable coincidence of views on growth, we can think of two celestial bodies on different orbits, momentarily coinciding from our earth-bound perspective at a single point in the sky and glowing with dazzling but transitory luminosity. The indelible image of this extraordinary event has been burned into the collective memory of economists, even if the details have long been forgotten. The resulting professional consensus, now obsolete, remained the guiding star for subsequent conceptual development and empirical observation for decades. 2.3.2. Human capital The initial challenge to the framework of Kuznets and Solow was posed by Denison’s magisterial study, Why Growth Rates Differ (1967). Denison retained NNP as a measure of national product and capital stock as a measure of capital input, adhering to the conventions employed by Kuznets and Solow. Denison’s comparisons among nine industrialized economies over the period 1950–1962 were cited extensively by both Kuznets and Solow. However, Denison departed from the identification of labor input with hours worked by Kuznets and Solow. He followed his earlier study of U.S. economic growth, The Sources of Economic Growth in the United States and the Alternatives Before Us, published in 1962 [Denison (1962)]. In this study he had constructed constant quality measures of labor input, taking into account differences in the quality of hours worked due to the age, sex, and educational attainment of workers. Kuznets (1971), recognizing the challenge implicit in Denison’s approach to measuring labor input, presented his own version of Denison’s findings.41 He carefully purged Denison’s measure of labor input of the effects of changes in educational attainment. Solow, for his part, made extensive references to Denison’s findings on the growth of output and capital stock, but avoided a detailed reference to Denison’s measure of labor input. Solow adhered instead to hours worked (or “man-hours” in the terminology of the early 1970s) as a measure of labor input.42 Kuznets showed that “. . . with one or two exceptions, the contribution of the factor inputs per capita was a minor fraction of the growth rate of per capita product”.43 For the United States during the period 1929 to 1957, the growth rate of productivity or output per unit of input exceeded the growth rate of output per capita. According to Kuznets’ estimates, the contribution of increases in capital input per capita over this extensive period was negative! 41 Kuznets (1971, Table 9, part B, pp. 74–75). 42 Solow (1970, pp. 2–7). However, Solow (1988, pp. 313–314), adopted Denison’s perspective on labor
input in his Nobel Prize address. At about the same time this view was endorsed by Gary Becker (1993a, p. 24), in his 1989 Ryerson Lecture at the University of Chicago. Becker (1993b) also cited Denison in his Nobel Prize address. 43 Kuznets (1971, p. 73).
Ch. 10:
Accounting for Growth in the Information Age
781
2.3.3. Solow’s surprise The starting point for our discussion of the demise of traditional growth accounting is a notable but neglected article by the great Dutch economist Jan Tinbergen (1959), published in German during World War II. Tinbergen analyzed the sources of U.S. economic growth over the period 1870–1914. He found that efficiency accounted only a little more than a quarter of growth in output, while growth in capital and labor inputs accounted for the remainder. This was precisely the opposite of the conclusion that Kuznets (1971) and Solow (1970) reached almost three decades later! The notion of efficiency or “total factor productivity” was introduced independently by George Stigler (1947) and became the starting point for a major research program at the National Bureau of Economic Research. This program employed data on output of the U.S. economy from earlier studies by the National Bureau, especially the pioneering estimates of the national product by Kuznets (1961). The input side employed data on capital from Raymond Goldsmith’s (1962) system of national wealth accounts. However, much of the data was generated by John Kendrick (1956, 1961), who employed an explicit system of national production accounts, including measures of output, input, and productivity for national aggregates and individual industries.44 The econometric models of Paul Douglas (1948) and Tinbergen were integrated with data from the aggregate production accounts generated by Abramovitz (1956) and Kendrick (1956) in Solow’s justly celebrated 1957 article, “Technical change and the aggregate production function” [Solow (1957)]. Solow identified “technical change” with shifts in the production function. Like Abramovitz, Kendrick, and Kuznets, he attributed almost all of U.S. economic growth to “residual” growth in productivity.45 Kuznets’ (1971) international comparisons strongly reinforced the findings of Abramovitz (1956), Kendrick (1956), and Solow (1957), which were limited to the United States.46 According to Kuznets, economic growth was largely attributable to the Solow residual between the growth of output and the growth of capital and labor inputs, although he did not use this terminology. Kuznets’ assessment of the significance of his empirical conclusions was unequivocal: (G)iven the assumptions of the accepted national economic accounting framework, and the basic demographic and institutional processes that control labor supply, capital accumulation, and initial capital-output ratios, this major conclusion – that the distinctive feature of modern economic growth, the high rate of growth of per
44 Updated estimates based on Kendrick’s framework are presented by Kendrick (1973) and Kendrick and
Eliot Grossman (1980). 45 This finding is called “Solow’s Surprise” by William Easterly (2001) and is listed as one of the “stylized
facts” about economic growth by Robert King and Sergio Rebelo (1999). 46 A survey of international comparisons, including Tinbergen (1959) and Kuznets (1971), is given in my
paper with Laurits Christensen and Dianne Cummings (1980), presented at the forty-fourth meeting of the Conference on Research and Wealth, held at Williamsburg, Virginia, in 1975.
782
D.W. Jorgenson
capita product is for the most part attributable to a high rate of growth in productivity – is inevitable.47 The empirical findings summarized by Kuznets have been repeatedly corroborated in investigations that employ the traditional approach to growth accounting. This approach identifies output with real NNP, labor input with hours worked, and capital input with real capital stock.48 Kuznets (1971) interpreted the Solow residual as due to exogenous technological innovation. This is consistent with Solow’s (1957) identification of the residual with technical change. Successful attempts to provide a more convincing explanation of the Solow residual have led, ultimately, to the demise of the traditional framework.49 2.3.4. Radical departure The most serious challenge to the traditional approach growth accounting was presented in my 1967 paper with Griliches, “The explanation of productivity change” [Jorgenson and Griliches (1967)]. Griliches and I departed far more radically than Denison from the measurement conventions of Kuznets and Solow. We replaced NNP with GNP as a measure of output and introduced constant quality indexes for both capital and labor inputs. The key idea underlying our constant quality index of labor input, like Denison’s, was to distinguish among different types of labor inputs. We combined hours worked for each type into a constant quality index of labor input, using the index number methodology Griliches (1960) had developed for U.S. agriculture. This considerably broadened the concept of substitution employed by Solow (1957). While he had modeled substitution between capital and labor inputs, Denison, Griliches and I extended the concept of substitution to include different types of labor inputs as well. This altered, irrevocably, the allocation of economic growth between substitution and technical change.50 Griliches and I introduced a constant quality index of capital input by distinguishing among types of capital inputs. To combine different types of capital into a constant quality index, we identified the prices of these inputs with rental prices, rather than the asset prices used in measuring capital stock. For this purpose we used a model of capital as a factor of production I had introduced in my 1963 article, “Capital theory and 47 Kuznets (1971, p. 73); see also, pp. 306–309. 48 For recent examples, see Michael Dertouzos, Solow and Richard Lester (1989) and Hall (1988, 1990a). 49 A detailed survey of research on sources of economic growth is given in my 1990 article, “Productivity
and economic growth” [Jorgenson (1990)], presented at the The Jubilee of the Conference on Research in Income and Wealth, held in Washington, DC, in 1988, commemorating the fiftieth anniversary of the founding of the Conference by Kuznets. More recent surveys are presented in Griliches’ (2000) posthumous book, R&D, Education, and Productivity, and Charles Hulten’s (2001) article, “Total factor productivity: A short biography”. 50 Constant quality indexes of labor input are discussed detail by Jorgenson, Gollop and Fraumeni (1987, Chapters 3 and 8, pp. 69–108 and 261–300), and Jorgenson, Ho and Stiroh (2004).
Ch. 10:
Accounting for Growth in the Information Age
783
investment behavior” [Jorgenson (1963)]. This made it possible to incorporate differences among depreciation rates on different assets, as well as variations in returns due to the tax treatment of different types of capital income, into our constant quality index of capital input.51 Finally, Griliches and I replaced the aggregate production function employed by Denison, Kuznets, and Solow with the production possibility frontier introduced in my 1966 paper, “The embodiment hypothesis” [Jorgenson (1966)]. This allowed for joint production of consumption and investment goods from capital and labor inputs. I had used this approach to generalize Solow’s (1960) concept of embodied technical change, showing that economic growth could be interpreted, equivalently, as “embodied” in investment or “disembodied” in productivity growth. My 1967 paper with Griliches [Jorgenson and Griliches (1967)] removed this indeterminacy by introducing constant quality price indexes for investment goods.52 Griliches and I showed that changes in the quality of capital and labor inputs and the quality of investment goods explained most of the Solow residual. We estimated that capital and labor inputs accounted for eighty-five percent of growth during the period 1945–1965, while only fifteen percent could be attributed to productivity growth. Changes in labor quality explained thirteen percent of growth, while changes in capital quality another eleven percent.53 Improvements in the quality of investment goods enhanced the growth of both investment goods output and capital input; the net contribution was only two percent of growth.54 2.3.5. The Rees Report The demise of the traditional framework for productivity measurement began with the Panel to Review Productivity Statistics of the National Research Council, chaired by Albert Rees. The Rees Report of 1979, Measurement and Interpretation of Productivity, 51 I have presented a detailed survey of empirical research on the measurement of capital input in my 1989
paper, “Capital as a factor of production [Jorgenson (1989)]”. Earlier surveys were given in my 1973 and 1980 papers [Jorgenson (1973, 1980)] and Diewert’s (1980) contribution to the forty-fifth meeting of the Conference on Income and Wealth, held at Toronto, Ontario, in 1976. Hulten (1990) surveyed conceptual aspects of capital measurement in his contribution to the Jubilee of the Conference on Research in Income and Wealth in 1988. 52 As a natural extension of Solow’s (1956) one-sector neo-classical model of economic growth, his 1960 model of embodiment had only a single output and did not allow for the introduction of a separate price index for investment goods. Recent research on Solow’s model of embodiment is surveyed by Greenwood and Boyan Jovanovic (2001) and discussed by Solow (2001). Solow’s model of embodiment is also employed by Karl Whelan (2002). 53 See Jorgenson and Griliches (1967, Table IX, p. 272). We also attributed thirteen percent of growth to the relative utilization of capital, measured by energy consumption as a proportion of capacity; however, this is inappropriate at the aggregate level, as Denison (1974, p. 56), pointed out. For additional details, see Jorgenson, Gollop and Fraumeni (1987), especially pp. 179–181. 54 Using Gordon’s (1990) estimates of improvements in the quality of producers’ durables, Hulten (1992) estimated this proportion as 8.5 percent of the growth of U.S. manufacturing output for the period 1949– 1983.
784
D.W. Jorgenson
became the cornerstone of a new measurement framework for the official productivity statistics. This was implemented by the Bureau of Labor Statistics (BLS), the U.S. government agency responsible for these statistics. Under the leadership of Jerome Mark and Edwin Dean the BLS Office of Productivity and Technology undertook the construction of a production account for the U.S. economy with measures of capital and labor inputs and total factor productivity, renamed multifactor productivity.55 The BLS (1983) framework was based on GNP rather than NNP and included a constant quality index of capital input, displacing two of the key conventions of the traditional framework of Kuznets and Solow.56 However, BLS retained hours worked as a measure of labor input until July 11, 1994, when it released a new multifactor productivity measure including a constant quality index of labor input as well. Meanwhile, BEA (1986) had incorporated a constant quality price index for computers into the national accounts – over the strenuous objections of Denison (1989). This index was incorporated into the BLS measure of output, completing the displacement of the traditional framework of economic measurement by the conventions employed in my papers with Griliches.57 The official BLS (1994) estimates of multifactor productivity have over-turned the findings of Abramovitz (1956) and Kendrick (1956), as well as those of Kuznets (1971) and Solow (1970). The official statistics have corroborated the findings summarized in my 1990 survey paper, “Productivity and economic growth” [Jorgenson (1990)]. These statistics are now consistent with the original findings of Tinbergen (1959), as well as my paper with Griliches (1967), and the results I have presented in Section 2.2. The approach to growth accounting presented in my 1987 book with Gollop and Fraumeni and the official statistics on multifactor productivity published by the BLS in 1994 has now been recognized as the international standard. The new framework for productivity measurement is presented in a Manual published by the Organization for Economic Co-Operation and Development (OECD) and written by Schreyer (2001). The expert advisory group for this manual was chaired by Dean, former Associate Commissioner for Productivity at the BLS, and leader of the successful effort to implement the Rees Report (1979).
3. International comparisons 3.1. Introduction In this section I present international comparisons of economic growth among the G7 nations – Canada, France, Germany, Italy, Japan, the United Kingdom, and the United 55 A detailed history of the BLS productivity measurement program is presented by Dean and Harper (2001). 56 The constant quality index of capital input became the international standard for measuring productivity
in Blades’ (2001) OECD manual, Measuring Capital. 57 The constant quality index of labor input became the international standard in the United Nations (1993)
System of National Accounts.
Ch. 10:
Accounting for Growth in the Information Age
785
States. These comparisons focus on the impact of investment in IT equipment and software over the period 1980–2001. In 1998 the G7 nations accounted for nearly 60 percent of world output58 and a much larger proportion of world investment in IT. Economic growth in the G7 has experienced a strong revival since 1995, driven by a powerful surge in IT investment. The resurgence of economic growth in the United States during the 1990s and the crucial role of IT investment has been thoroughly documented and widely discussed.59 Similar trends in the other G7 economies have been more difficult to detect, partly because of discrepancies among official price indexes for IT equipment and software identified by Andrew Wyckoff (1995).60 Paul Schreyer (2000) has constructed “internationally harmonized” IT prices that eliminate many of these discrepancies.61 Using internationally harmonized prices, I have analyzed the role of investment and total factor productivity as sources of growth in the G7 countries over the period 1980– 2001. I have subdivided the period in 1989 and 1995 in order to focus on the most recent experience, particularly the post-1995 surge. I have decomposed growth of output for each country between growth of input and total factor productivity. Finally, I have allocated the growth of input between investments in tangible assets, especially information technology and software, and human capital. Growth in IT capital input per capita jumped to double-digit levels in the G7 nations after 1995. This can be traced to acceleration in the rate of decline of IT prices, analyzed in my Presidential Address to the American Economic Association.62 The powerful surge in investment was most pronounced in Canada, but capital input growth in the United States, the United Kingdom, and Japan was only slightly lower. France, Germany, and Italy also experienced double-digit growth, but lagged considerably behind the leaders. During the 1980s total factor productivity played a minor role as a source of growth for the G7 countries except Japan, where total factor productivity accounted for 25 percent of economic growth. Total factor productivity accounted for only 15 percent of growth in the United States, 13 percent in France, 12 percent in the United Kingdom, and 11 percent in Germany; only 2 percent of growth in Canada was due to total factor productivity, while the decline of total factor productivity retarded growth by 14 percent in Italy. Between 1989 and 1995 total factor productivity growth declined further in the G7 nations, except for Italy and Germany. Total factor productivity declined for France and the United Kingdom but remained positive for the United States, Canada, and Japan. Total factor productivity growth revived in all the G7 countries after 1995, with the exception of Germany and Italy, where it declined, and Japan, where it remained 58 See Angus Maddison (2001) for 1998 data for world GDP and the GDP of each of the G7 countries. 59 See Jorgenson and Stiroh (2000b) and Oliner and Sichel (2000). 60 See Wyckoff (1995). 61 See Schreyer (2000). Alessandra Colecchia and Schreyer (2002) have employed these internationally har-
monized prices in measuring the impact of IT investment. 62 See Jorgenson (2001).
786
D.W. Jorgenson
very similar. The resurgence was most dramatic in Canada, The United Kingdom, and France, partly offsetting years of dismal total factor productivity growth. Japan exhibited the highest growth in output per capita among the G7 nations from 1980 to 1995. Japan’s level of output per capita was the lowest in the G7. Japan’s total factor productivity growth far outstripped the other members of the G7 with the exception of the United Kingdom between 1995–2001. Nonetheless, Japan’s total factor productivity remained the lowest among the G7 nations. The United States led the G7 in output per capita for the period 1989–2000. Canada’s edge in output per capita in 1980 had disappeared by 1989. The United States led the G7 countries in input per capita during 1980–2000, but U.S. total factor productivity languished below the levels of Canada, France, and Italy. In Section 3.2 I outline the methodology for this study, based on Section 2. I have utilized the U.S. data presented there through 2001. Comparable data on investment in information technology have been constructed for Canada by Statistics Canada.63 Data on information technology for France, Germany, Italy, and the United Kingdom have been developed for the European Commission by Bart Van Ark et al.64 Finally, data for Japan have been assembled by myself and Kazuyuki Motohashi for the Research Institute on Economy, Trade, and Industry.65 I have linked these data by means of the OECD’s purchasing power parities for 1999.66 In Section 3.3 I consider the impact of IT investment and the relative importance of investment and total factor productivity in accounting for economic growth among the G7 nations. Investments in human capital and tangible assets, especially IT equipment and software, account for the overwhelming proportion of growth. Differences in the composition of capital and labor inputs are essential for identifying persistent international differences in output and accounting for the impact of IT investment. In Section 3.4 I consider alternative approaches to international comparisons. The great revival of interest in economic growth among economists dates from Maddison’s (1982) updating and extension of Simon Kuznets’ (1971) long-term estimates of the growth of national product and population for 14 industrialized countries, including the G7 nations. Maddison (1982, 1991) added Austria and Finland to Kuznets’ list and presented growth rates covering periods beginning as early as 1820 and extending through 1989. Maddison (1987, 1991) also generated growth accounts for major industrialized countries, but did not make level comparisons like those presented in Section 3.2 below. As a consequence, total factor productivity differences were omitted from the canonical formulation of “growth regressions” by William Baumol (1986). This proved to be a fatal flaw in Baumol’s regression model, remedied by Nazrul Islam’s (1995) panel data model. Section 3.5 concludes. 63 See John Baldwin and Tarek Harchaoui (2002). 64 See Van Ark et al. (2002). 65 See Jorgenson and Motohashi (2003). 66 See OECD (2002). Current data on purchasing power parities are available from the OECD website:
http://www.sourceoecd.org.
Ch. 10:
Accounting for Growth in the Information Age
787
3.2. Investment and total factor productivity My papers with Laurits Christensen and Dianne Cummings (1980, 1981) developed growth accounts for the United States and its major trading partners – Canada, France, Germany, Italy, Japan, Korea, The Netherlands, and the United Kingdom for 1947– 1973. We employed GNP as a measure of output and incorporated constant quality indices of capital and labor input for each country. Our 1981 paper compared levels of output, inputs, and total factor productivity for all nine nations. I have updated the estimates for the G7 – Canada, France, Germany, Italy, Japan, the United Kingdom, and the United States – through 1995 in earlier work. The updated estimates are presented in my papers with Chrys Dougherty (1996, 1997) and Eric Yip (2000). We have shown that total factor productivity accounted for only 11 percent of economic growth in Canada and the United States over the period 1960–1995. My paper with [Jorgenson and Yip (2000)] attributed 47 percent of Japanese economic growth during the period 1960–1995 to total factor productivity growth. The proportion attributable to total factor productivity approximated 40 percent of growth for the four European countries – France (38 percent), Germany (42 percent), Italy (43 percent), and the United Kingdom (36 percent). Input growth predominated over productivity growth for all the G7 nations. I have now incorporated new data on investment in information technology equipment and software for the G7. I have also employed internationally harmonized prices like those constructed by Schreyer (2000). As a consequence, I have been able to separate the contribution of capital input to economic growth into IT and non-IT components. While IT investment follows similar patterns in all the G7 nations, non-IT investment varies considerably and helps to explain important differences in growth rates among the G7. 3.2.1. Comparisons of output, input, and productivity My first objective is to extend my estimates for the G7 nations with Christensen, Cummings, Dougherty, and Yip to the year 2001. Following the methodology of my Presidential Address, I have chosen GDP as a measure of output. I have included imputations for the services of consumers’ durables as well as land, buildings, and equipment owned by nonprofit institutions. I have also distinguished between investments in IT equipment and software and investments in other forms of tangible assets. A constant quality index of capital input is based on weights that reflect differences in capital consumption, tax treatment, and the rate of decline of asset prices. I have derived estimates of capital input and property income from national accounting data. Similarly, a constant quality index of labor input is based on weights by age, sex, educational attainment, and employment status. I have constructed estimates of hours worked and labor compensation from labor force surveys for each country. In Table 9 I present output per capita for the G7 nations from 1980 to 2001, taking the United States as 100.0 in 2000. Output and population estimates are given separately in
788
D.W. Jorgenson Table 9 Levels of output and input per capita and total factor productivity
Year
U.S.
Canada
UK
1980 1989 1995 2001
63.9 79.7 85.6 100.3
67.6 78.8 79.6 91.9
45.0 56.5 61.4 71.3
1980 1989 1995 2001
70.5 83.9 88.8 100.8
64.2 74.4 75.2 83.7
50.2 61.2 67.0 73.6
1980 1989 1995 2001
90.6 94.9 96.4 99.5
105.4 105.9 105.9 109.7
89.5 92.3 91.7 96.9
France
Germany
Output per capita 45.9 49.3 54.1 58.6 57.0 65.0 64.0 69.2 Input per capita 46.5 61.0 53.3 71.1 57.0 73.7 61.7 79.0 Total factor productivity 98.6 80.8 101.5 82.4 99.9 88.1 103.6 87.6
Italy
Japan
45.9 57.3 62.1 68.8
37.5 50.9 57.5 63.6
43.1 55.5 58.8 67.2
52.3 64.8 69.8 73.3
106.6 103.2 105.6 102.5
71.7 78.5 82.5 86.8
Note: U.S. = 100.0 in 2000, Canada data begins in 1981.
Tables 10 and 11, respectively. I use 1999 purchasing power parities from the OECD to convert output from domestic prices for each country into U.S. dollars. The United States gained the lead among the G7 countries in output per capita after 1989. Canada led the United States in 1980, but fell behind in 1989. The U.S.–Canada gap widened considerably during the 1990s. The four major European nations – the United Kingdom, France, Germany, and Italy – had very similar levels of output per capita throughout the period 1980–1989. Japan remained in last place from 1980 to 2001, lagging considerably behind the United States and Canada, but only slightly behind France in 2001. Japan led the G7 in the growth of output per capita from 1980–1989, but fell behind all but Germany after 1995. In Table 9 I present input per capita for the G7 over the period 1980–2001, taking the United States as 100.0 in 2000. I express input per capita in U.S. dollars, using purchasing power parities constructed for this study.67 The United States was the leader among the G7 in input per capita throughout the period. In 2001 Canada ranked next to the United States with Germany third and the United Kingdom and Japan close behind. France and Italy started at the bottom of the ranking and remained there throughout the period.
67 The purchasing power parities for outputs are based on OECD (2002). Purchasing power parities for inputs
follow the methodology described in detail by Jorgenson and Yip (2000).
Ch. 10:
Accounting for Growth in the Information Age
789
Table 10 Growth rate and level of output Year
1980–1989 1989–1995 1995–2001
U.S.
Canada
3.38 2.43 3.76
3.10 1.39 3.34
1980 1989 1995 2001
5361.2 7264.2 8403.3 10530.4
618.4 792.6 861.4 1052.3
1980 1989 1995 2001
51.6 69.9 80.8 101.3
5.9 7.6 8.3 10.1
UK
France
Germany
Growth rate (percentage) 2.69 2.38 1.99 1.62 1.30 2.34 2.74 2.34 1.18 Level (billions of 2000 U.S. dollars) 934.0 932.0 1421.7 1190.3 1154.3 1700.2 1311.8 1247.8 1956.3 1545.9 1436.0 2099.8 Level (U.S. = 100.0 in 2000) 9.0 9.0 13.7 11.4 11.1 16.3 12.6 12.0 18.8 14.9 13.8 20.2
Italy
2.51 1.52 1.90
Japan
3.98 2.39 1.89
955.7 1197.4 1311.5 1470.1
1612.9 2308.3 2663.7 2983.3
9.2 11.5 12.6 14.1
15.5 22.2 25.6 28.7
Italy
Japan
0.05 0.18 0.18
0.59 0.33 0.22
Note: Canada data begins in 1981.
Table 11 Growth rate and level in population Year
1980–1989 1989–1995 1995–2001
U.S.
0.92 1.23 1.12
Canada
1.18 1.22 0.95
1980 1989 1995 2001
227.7 247.4 266.3 284.8
24.8 27.3 29.4 31.1
1980 1989 1995 2001
80.7 87.7 94.4 101.0
8.8 9.7 10.4 11.0
UK
France
Germany
Growth rate 0.54 0.05 0.45 0.62 0.41 0.14 Level (millions) 56.3 55.1 78.3 57.1 57.9 78.7 58.0 59.4 81.7 58.8 60.9 82.3 Level (U.S. = 100.0 in 2000) 20.0 19.5 27.8 20.3 20.5 27.9 20.5 21.1 28.9 20.8 21.6 29.2 0.16 0.24 0.24
56.4 56.7 57.3 57.9
116.8 123.1 125.6 127.2
20.0 20.1 20.3 20.5
41.4 43.6 44.5 45.1
Note: Percentage, Canada data begins in 1981.
Finally, Table 9 presents total factor productivity levels for the G7 over the period 1980–2001. Total factor productivity is defined as the ratio of output to input, including both capital and labor inputs. Italy led in 1980 and Canada was the total factor
790
D.W. Jorgenson Table 12 Growth in output and input per capita and total factor productivity
Year
U.S.
Canada
1980–1989 1989–1995 1995–2001
2.46 1.20 2.64
1.92 0.17 2.38
1980–1989 1989–1995 1995–2001
1.94 0.94 2.10
1.86 0.17 1.80
1980–1989 1989–1995 1995–2001
0.52 0.26 0.54
0.06 0.00 0.58
UK
France
Germany
Output per capita 1.84 1.93 0.85 1.72 1.93 1.04 Input per capita 2.20 1.52 1.71 1.49 1.11 0.60 1.59 1.33 1.14 Total factor productivity 0.34 0.32 0.23 −0.11 −0.26 1.12 0.91 0.60 −0.10 2.54 1.38 2.50
Italy
Japan
2.46 1.33 1.72
3.40 2.06 1.67
2.82 0.96 2.21
2.38 1.23 0.83
−0.36 0.37 −0.49
1.01 0.83 0.85
Note: Percentage, Canada data begins in 1981.
productivity leader throughout the period 1989–2001 with France close behind. Japan made substantial gains in total factor productivity during the period, while there were more modest increases in the United States, Canada, the United Kingdom, France, and Germany, and a decline in Italy. I summarize growth in output and input per capita and total factor productivity for the G7 nations in Table 12, while the growth rates of output and population for the period 1980–2001 in Tables 10 and 11. Output growth slowed in the G7 after 1989, but revived for all nations except Japan and Germany after 1995. Output per capita followed a similar pattern with Canada barely expanding during the period 1989–1995. Japan led in growth of output in the 1980s, and output per capita through 1995, but fell to the lower echelon of the G7 after 1995. Japan also led in total factor productivity growth during the period 1980–1995. For all countries and all time periods, except for Germany during the period 1989–1995 and Japan after 1995, the growth of input per capita exceeded growth of total factor productivity by a substantial margin. Total factor productivity growth in the G7 slowed during the period 1989–1995, except for Germany and Italy, where total factor productivity slumped after 1995. Italy led the G7 in growth of input per capita for the periods 1980–1989 and 1995– 2001, but relinquished leadership to the United Kingdom for the period 1989–1995. Differences among input growth rates were smaller than differences among output growth rates, but there was a slowdown in input growth during 1989–1995 throughout the G7. After 1995 growth of input per capita increased in every G7 nation except Japan.
Ch. 10:
Accounting for Growth in the Information Age
791
Table 13 Levels of capital input and capital stock per capita and capital quality Year
U.S.
Canada
1980 1989 1995 2001
57.7 73.7 81.6 103.9
56.0 67.1 68.3 78.0
1980 1989 1995 2001
76.8 88.4 92.2 101.7
40.7 48.5 50.8 55.1
1980 1989 1995 2001
75.1 83.4 88.5 102.2
137.5 138.2 134.6 141.5
UK
France
Germany
Capital input per capita 36.3 48.3 52.7 58.1 Capital stock per capita 24.1 36.2 31.2 42.4 35.9 47.0 44.5 52.0 Capital quality 107.0 100.1 121.7 114.0 139.3 112.2 126.1 111.9 25.8 37.9 50.0 56.1
Italy
Japan
44.6 62.1 72.3 83.5
35.6 62.4 73.1 89.4
25.2 35.0 43.2 52.4
60.2 67.9 77.0 85.5
36.0 52.4 62.3 72.3
85.6 91.5 97.0 100.2
74.0 91.5 94.0 97.7
98.8 119.1 117.4 123.6
29.4 38.3 44.5 52.3
Note: U.S. = 100.0 in 2000, Canada data begins in 1981.
3.2.2. Comparisons of capital and labor quality A constant quality index of capital input weights capital inputs by property compensation per unit of capital. By contrast an index of capital stock weights different types of capital by asset prices. The ratio of capital input to capital stock measures the average quality of a unit of capital. This represents the difference between the constant quality index of capital input and the index of capital stock employed, for example, by Kuznets (1971) and Solow (1970). In Table 13 I present capital input per capita for the G7 countries over the period 1980–2001 relative to the United States in 2000. The United States was the leader in capital input per capita throughout the period, while Japan was the laggard. Canada led the remaining six countries in 1980, but was overtaken by Germany and Italy in 1995. Italy led the rest of the G7 through 2001, but lagged considerably behind the United States. The picture for capital stock per capita has some similarities to capital input, but there are important differences. Capital stock levels do not accurately reflect the substitutions among capital inputs that accompany investments in tangible assets, especially investments in IT equipment and software. Japan led the G7 in capital stock per capita until 2001, when the United States took the lead. The United Kingdom lagged the remaining countries of the G7 throughout the period. The behavior of capital quality highlights the differences between the constant quality index of capital input and capital stock. There are important changes in capital quality over time and persistent differences among countries, so that heterogeneity in capital in-
792
D.W. Jorgenson Table 14 Levels of IT capital input and IT capital stock per capita and IT capital quality
Year
U.S.
Canada
1980 1989 1995 2001
4.5 19.3 38.1 115.3
1.0 3.9 11.2 45.6
1980 1989 1995 2001
9.8 27.4 46.8 110.7
0.8 3.7 9.7 31.8
1980 1989 1995 2001
46.4 70.4 81.3 104.1
118.4 107.4 115.0 143.4
UK
France
Germany
IT capital input per capita 3.0 4.2 7.1 10.9 11.9 18.7 20.9 19.1 31.1 53.6 38.1 59.7 IT capital stock per capita 2.5 3.5 6.1 9.6 9.9 15.5 19.2 18.0 28.2 44.9 33.4 49.7 IT capital quality 118.5 117.5 117.4 112.7 119.7 120.4 108.9 106.2 110.1 119.3 114.1 120.2
Italy
Japan
6.7 18.8 31.2 60.3
2.3 12.1 21.9 55.5
4.6 13.1 23.8 44.1
4.5 17.5 28.7 73.1
146.8 143.2 131.0 136.6
51.7 69.1 76.2 75.9
Note: U.S. = 100.0 in 2000, Canada data begins in 1981.
put must be taken into account in international comparisons of economic performance. Throughout the period 1980–2001, Canada led the G7 in capital quality, with the exception of the United Kingdom, which edged ahead in the mid 1990s. Japan remained at the bottom. I summarize growth in capital input and capital stock per capita, as well as capital quality for the G7 nations in Table 16. Italy was the international leader in capital input growth from 1980–1989, while Canada was the laggard. The United Kingdom led from 1989–1995, while Canada lagged considerably behind the rest of the G7. The United States took the lead after 1995. There was a slowdown in capital input growth throughout the G7 after 1989, except for the United Kingdom, and a revival after 1995 in the United States, Canada, France, and Italy. A constant quality index of labor input weights hours worked for different categories by labor compensation per hour. An index of hours worked fails to take quality differences into account. The ratio of labor input to hours worked measures the average quality of an hour of labor, as reflected in its marginal product. This represents the difference between the constant quality index of labor input and the index of hours worked employed, for example, by Kuznets (1971) and Solow (1970). In Table 19 I present labor input per capita for the G7 nations for the period 1980– 2001 relative to the United States in 2000. Japan was the international leader for the period 1980–1995 with the United States leading in 2001, and France and Italy the laggards. Labor input in Japan was nearly double that of Italy. The United States led the remaining G7 nations throughout the period. The United Kingdom ranked third among
Ch. 10:
Accounting for Growth in the Information Age
793
Table 15 Levels of non-IT capital input and capital stock per capita and non-IT capital quality Year
U.S.
Canada
1980 1989 1995 2001
73.8 87.0 90.7 102.2
73.1 83.1 79.9 84.0
1980 1989 1995 2001
82.5 92.5 94.8 101.4
44.1 51.5 53.0 57.4
1980 1989 1995 2001
89.5 94.1 95.6 100.8
165.7 161.2 150.7 146.5
UK
France
Germany
Non-IT capital input per capita 30.7 41.3 51.9 43.4 53.9 70.3 55.9 57.9 79.7 56.4 62.6 87.3 Non-IT capital stock per capita 25.7 38.0 63.4 32.6 44.0 70.6 36.9 48.3 79.3 44.5 54.1 87.2 Non-IT capital quality 119.2 108.5 81.9 133.2 122.6 99.5 151.5 119.9 100.5 126.7 115.8 100.1
Italy
Japan
41.6 71.3 81.2 94.7
31.3 39.9 47.4 48.3
38.2 54.8 64.4 75.1
91.6 96.6 101.6 102.5
109.2 130.0 126.0 126.1
34.2 41.3 46.6 47.1
Note: U.S. = 100.0 in 2000, Canada data begins in 1981. Table 16 Growth in capital input and capital stock per capita and capital quality Year
U.S.
Canada
1980–1989 1989–1995 1995–2001
2.72 1.70 4.03
2.26 0.31 2.20
1980–1989 1989–1995 1995–2001
1.56 0.70 1.63
2.19 1.05 1.36
1980–1989 1989–1995 1995–2001
1.17 0.99 2.40
0.07 −0.74 0.84
UK
France
Germany
Capital input per capita 3.19 3.70 1.46 2.53 1.63 2.40 Capital stock per capita 2.85 1.74 1.34 2.36 1.74 2.09 3.57 1.67 1.75 Capital quality 1.43 1.45 2.36 2.25 −0.27 0.44 −1.65 −0.04 0.65 4.28 4.61 1.92
Italy
Japan
6.25 2.63 3.35
3.67 3.49 3.21
4.18 2.87 2.49
0.74 0.97 0.53
2.07 −0.24 0.86
2.93 2.51 2.68
Note: Percentage, Canada data begins in 1981.
the G7 through 1995. Italy and France lagged behind the rest of the G7 for the entire period. The picture for hours worked per capita has some similarities to labor input, but there are important differences. Japan was the international leader in hours worked per capita. The United States, Canada, and the United Kingdom moved roughly in parallel.
794
D.W. Jorgenson Table 17 Growth in IT capital input and capital stock per capita and IT capital quality
Year
U.S.
Canada
1980–1989 1989–1995 1995–2001
16.09 11.35 18.47
17.66 17.42 23.42
1980–1989 1989–1995 1995–2001
11.47 8.94 14.34
18.88 16.28 19.73
1980–1989 1989–1995 1995–2001
4.63 2.41 4.12
−1.22 1.14 3.69
UK
France
Germany
IT capital input per capita 14.43 11.66 10.71 10.91 7.92 8.47 15.69 11.55 10.87 IT capital stock per capita 14.98 11.46 10.43 11.50 9.91 9.97 14.16 10.35 9.40 IT capital quality −0.56 0.20 0.28 −0.58 −1.99 −1.50 1.53 1.20 1.47
Italy
Japan
11.44 8.44 10.98
18.33 9.92 15.49
11.72 9.94 10.28
15.10 8.29 15.55
−0.27 −1.49 0.70
3.23 1.63 −0.06
Note: Percentage, Canada data begins in 1981.
Table 18 Growth in non-IT capital input and capital stock per capita and non-IT capital quality Year
U.S.
Canada
1980–1989 1989–1995 1995–2001
1.83 0.68 2.00
1.60 −0.66 0.85
1980–1989 1989–1995 1995–2001
1.27 0.41 1.11
1.94 0.47 1.32
1980–1989 1989–1995 1995–2001
0.56 0.27 0.88
−0.35 −1.13 −0.47
UK
France
Germany
Non-IT capital input per capita 3.85 2.97 3.36 4.22 1.20 2.09 0.15 1.30 1.52 Non-IT capital stock per capita 2.62 1.61 1.20 2.07 1.58 1.92 3.12 1.87 1.59 Non-IT capital quality 1.23 1.36 2.16 2.15 −0.38 0.17 −2.97 −0.57 −0.06
Italy
Japan
5.97 2.17 2.57
2.69 2.85 0.33
4.03 2.68 2.56
0.59 0.84 0.15
1.94 −0.51 0.01
2.10 2.01 0.18
Note: Percentage, Canada data begins in 1981.
The United Kingdom ranked second in 1980 and 1989, while the United States ranked second in 1995 and 2001. France and Italy lagged the rest of the G7 from 1980–2001. The behavior of labor quality highlights the differences between labor input and hours worked. Germany was the leader in labor quality throughout the period 1980–2001 with the United States close behind. Canada, the United Kingdom, France, and Japan had similar levels of labor quality throughout the period, but fell short of German and U.S. levels. Italy was the laggard among the G7 in labor quality.
Ch. 10:
Accounting for Growth in the Information Age
795
Table 19 Levels of labor input and hours worked per capita and labor quality Year
U.S.
Canada
1980 1989 1995 2001
81.1 91.9 94.2 98.8
73.0 82.1 82.3 89.3
1980 1989 1995 2001
89.7 97.1 95.9 98.3
91.4 96.6 90.9 96.3
1980 1989 1995 2001
90.4 94.7 98.2 100.5
79.9 85.0 90.6 92.7
UK
France
Germany
Labor input per capita 63.0 75.4 59.4 78.7 61.7 75.2 65.3 75.9 Hours worked per capita 92.0 79.3 82.3 97.7 71.2 82.7 89.8 67.6 76.4 94.2 69.7 75.3 Labor quality 85.7 79.5 91.6 87.4 83.5 95.2 91.7 91.2 98.4 94.7 93.7 100.9 78.9 85.4 82.4 89.2
Italy
Japan
48.8 51.0 50.6 55.1
84.8 97.4 95.6 91.4
71.4 72.1 68.9 72.3
111.9 115.6 109.9 101.3
68.3 70.7 73.5 76.1
75.8 84.3 87.0 90.3
Italy
Japan
0.49 −0.13 1.40
1.53 −0.31 −0.74
0.10 −0.75 0.81
0.36 −0.84 −1.36
0.39 0.63 0.60
1.18 0.52 0.62
Note: U.S. = 100.0 in 2000, Canada data begins in 1981.
Table 20 Growth in labor input and hours worked per capita and labor quality Year
U.S.
Canada
1980–1989 1989–1995 1995–2001
1.38 0.41 0.79
1.47 0.04 1.35
1980–1989 1989–1995 1995–2001
0.87 −0.21 0.41
0.69 −1.02 0.98
1980–1989 1989–1995 1995–2001
0.51 0.61 0.38
0.78 1.06 0.38
UK
France
Germany
Labor input per capita 0.88 −0.65 0.48 −0.59 0.61 −0.78 1.32 0.95 0.17 Hours worked per capita 0.67 −1.20 0.06 −1.41 −0.86 −1.33 0.79 0.50 −0.25 Labor quality 0.21 0.55 0.42 0.81 1.47 0.55 0.53 0.45 0.41
Note: Percentage, Canada data begins in 1981.
I summarize growth in labor input and hours worked per capita, as well as labor quality for the period 1980–2001 in Table 20. Canada and Japan led the G7 nations in labor input growth during the 1980s, France led from 1989–1995 but relinquished its leadership to Italy after 1995. Labor input growth was negative for France during the
796
D.W. Jorgenson
1980s, for the United Kingdom, Germany, and Italy during the period 1989–1995, and for Japan after 1989. Hours worked per capita fell continuously throughout the period 1980–2001 for Japan and declined for all the G7 nations during the period 1989–1995. Growth in labor quality was positive for the G7 nations in all time periods. Japan was the leader during the 1980s, relinquishing its lead to France during the early 1990s and taking back the lead in the late 1990s. Growth in labor quality and hours worked are equally important as sources of growth in labor input for the G7. 3.2.3. The relative importance of investment and total factor productivity Using data from Tables 9 and 10, I can assess the relative importance of investment and total factor productivity as sources of economic growth for the G7 nations. The main conclusion is that investments in tangible assets and human capital greatly predominated over total factor productivity during the period 1980–2001. While total factor productivity fell in Italy during this period, the remaining G7 countries had positive total factor productivity growth for the period as a whole. Similarly, using data from Table 13 I can assess the relative importance of growth in capital stock and capital quality. Capital input growth was positive for all countries for the period 1980–2001 and all three sub-periods. Capital quality growth was positive for the period as a whole for all G7 countries. Although capital stock predominated in capital input growth, capital quality was also quantitatively significant, especially after 1995. Finally, using data from Table 19 I can assess the relative importance of growth in hours worked and labor quality. Hours worked per capita declined for France, Germany, and Japan during the periods 1980–1989 and 1995–2001, while labor quality rose in these nations during the period 1980–2001. For the United States, Canada, the United Kingdom, and Italy, both hours worked per capita and labor quality rose. I conclude that labor quality growth is essential to the analysis of growth in labor input. 3.3. Investment in information technology The final step in the comparison of patterns of economic growth among the G7 nations is to analyze the impact of investment in information technology equipment and software. In Table 14 I present levels of IT capital input per capita for the G7 for the period 1980– 2001, relative to the United States in 2000. The United States overtook Germany in 1989 and remained the leader through 2001. Canada and Japan lagged behind the rest of the G7 through 1995, but France fell into last place in 2001. Table 14 reveals substantial differences between IT capital stock and IT capital input. The G7 nations began with very modest stocks of IT equipment and software per capita in 1980. These stocks expanded rapidly during the period 1980–2001. The United States led in IT capital stock throughout the period, while Japan moved from the third lowest level in 1980 to the second highest from 1989–2001.
Ch. 10:
Accounting for Growth in the Information Age
797
IT capital quality reflects differences in the composition of IT capital input, relative to IT capital stock. A rising level of capital quality indicates a shift toward short-lived assets, such as computers and software. This shift is particularly dramatic for the United States, Canada and Japan, while the composition of IT capital stock changed relatively less for the United Kingdom, France, Germany, and Italy. Patterns for non-IT capital input, capital stock, and capital quality in Table 15 largely reflect those for capital as a whole, presented in Table 13. I present growth rates for IT capital input per capita, capital stock per capita, and capital quality in Table 17. The G7 nations have exhibited double-digit growth in IT capital input per capita since 1995. Canada was the international leader during this period with the United States close behind. Japan was the leader in growth of IT capital input during the 1980s, another period of double-digit growth in the G7. However, Japanese IT growth slowed substantially during 1989–1995 and Canada gained the lead. Patterns of growth for IT capital stock per capita are similar to those for IT capital input for the four European countries. Changes in the composition of IT capital stock per capita were important sources of growth of IT capital input per capita for the United States, Canada, and Japan. IT capital stock also followed the pattern of IT capital input with substantial growth during the 1980s, followed by a pronounced lull during the period 1989–1995. After 1995 the growth rates of IT capital stock surged in all the G7 countries, except Germany, but exceeded the rates of the 1980s only for the United States and Japan. Finally, growth rates for IT capital quality reflect the rates at which shorter-lived IT assets are substituted for longer-lived assets. The United States led in the growth of capital quality throughout the period. IT capital quality growth for the United States outstripped that of the remaining G7 countries for the period 1980–2001. Patterns of growth in non-IT capital input per capita, non-IT capital stock per capita, and non-IT capital quality given in Table 18 largely reflect those for capital as a whole presented in Table 16. Table 21 and Figure 13 present the contribution of capital input to economic growth for the G7 nations, divided between IT and non-IT. The powerful surge of IT investment in the United States after 1995 is mirrored in similar jumps in growth rates of the contribution of IT capital through the G7. The contribution of IT capital input was similar during the 1980s and the period 1989–1995 for all the G7 nations, despite the dip in rates of economic growth after 1989. Japan is an exception to this general pattern with a contribution of IT capital comparable to that of the United States during the 1980s, followed by a decline in this contribution from 1989–1995, reflecting the sharp downturn in Japanese economic growth. The contribution of non-IT capital input to economic growth after 1995 exceeded that for IT capital input for four of the G7 nations; the exceptions were Canada, the United Kingdom, and Japan. The United States stands out in the magnitude of the contribution of capital input after 1995. Both IT and non-IT capital input contributed to the U.S. economic resurgence of the last half of the 1990s. Despite the strong performance of IT
798
D.W. Jorgenson Table 21 Contribution of total capital, IT capital and non-IT capital to output growth
Year
U.S.
Canada
UK
1980–1989 1989–1995 1995–2001
1.53 1.19 2.10
1.71 0.76 1.67
1.80 1.96 0.94
1980–1989 1989–1995 1995–2001
0.45 0.49 0.99
0.39 0.49 0.86
0.24 0.27 0.76
1980–1989 1989–1995 1995–2001
1.08 0.70 1.11
1.32 0.27 0.81
1.56 1.69 0.18
France Total capital 2.12 1.12 1.15 IT capital 0.18 0.19 0.42 Non-IT capital 1.94 0.93 0.73
Germany
Italy
Japan
1.44 1.31 1.11
2.55 1.12 1.47
1.68 1.56 1.36
0.19 0.26 0.46
0.24 0.26 0.49
0.47 0.37 0.79
1.25 1.05 0.65
2.31 0.86 0.98
1.21 1.19 0.57
Note: Percentage. Contribution is growth rate times value share. Canada data begins in 1981.
Figure 13. Capital input contribution by country.
investment in Japan after 1995, the contribution of capital input declined substantially; the pattern for the United Kingdom is similar. Table 22 and Figure 14 present contributions to economic growth from total factor productivity, divided between the IT-producing and non-IT-producing industries. The
Ch. 10:
Accounting for Growth in the Information Age
799
Table 22 Contributions of productivity from IT and non-IT production to output growth Year
U.S.
Canada
1980–1989 1989–1995 1995–2001
0.52 0.26 0.54
0.06 0.00 0.58
1980–1989 1989–1995 1995–2001
0.23 0.23 0.48
0.14 0.14 0.17
1980–1989 1989–1995 1995–2001
0.29 0.03 0.06
−0.08 −0.14 0.41
UK
France
Germany
Productivity 0.34 0.32 0.23 −0.11 −0.26 1.12 0.91 0.60 −0.10 Productivity from IT production 0.23 0.29 0.28 0.32 0.29 0.43 0.82 0.56 0.65 Productivity from non-IT production 0.11 0.03 −0.05 −0.43 −0.55 0.69 0.09 0.04 −0.75
Italy
Japan
−0.36 0.37 −0.49
1.01 0.83 0.85
0.32 0.38 0.68
0.19 0.20 0.43
−0.68 −0.01 −1.17
0.82 0.63 0.42
Note: Percentage. Canada data begins in 1981.
Figure 14. Sources of total factor productivity growth by country.
methodology for this division follows Triplett (1996). The contribution of IT-producing industries is positive throughout the period 1980–2001 and jumps substantially after 1995. Since the level of total factor productivity in Italy is higher in 1980 than in 2001, it is not surprising that the contribution of total factor productivity growth in the non-IT
800
D.W. Jorgenson Table 23 Sources of output growth
Year
U.S.
Canada
1980–1989 1989–1995 1995–2001
3.38 2.43 3.76
3.10 1.39 3.34
1980–1989 1989–1995 1995–2001
1.33 0.98 1.12
1.33 0.62 1.08
1980–1989 1989–1995 1995–2001
0.45 0.49 0.99
0.39 0.49 0.86
1980–1989 1989–1995 1995–2001
1.08 0.70 1.11
1.32 0.27 0.81
1980–1989 1989–1995 1995–2001
0.23 0.23 0.48
0.14 0.14 0.17
1980–1989 1989–1995 1995–2001
0.29 0.03 0.06
−0.08 −0.14 0.41
UK
France
Germany
Output 2.38 1.99 1.30 2.34 2.34 1.18 Labor 0.56 −0.06 0.32 −0.24 0.44 −0.09 0.88 0.59 0.17 IT capital 0.24 0.18 0.19 0.27 0.19 0.26 0.76 0.42 0.46 Non-IT capital 1.56 1.94 1.25 1.69 0.93 1.05 0.18 0.73 0.65 Productivity from IT production 0.23 0.29 0.28 0.32 0.29 0.43 0.82 0.56 0.65 Productivity from non-IT production 0.11 0.03 −0.05 −0.43 −0.55 0.69 0.09 0.04 −0.75 2.69 1.62 2.74
Italy
Japan
2.51 1.52 1.90
3.98 2.39 1.89
0.32 0.03 0.93
1.29 0.00 −0.32
0.24 0.26 0.49
0.47 0.37 0.79
2.31 0.86 0.98
1.21 1.19 0.57
0.32 0.38 0.68
0.19 0.20 0.43
−0.68 −0.01 −1.17
0.82 0.63 0.42
Note: Percentage. Contributions. Canada data begins in 1981.
industries was negative throughout the period. Total factor productivity in these industries declined after 1989 in the United Kingdom, France and Germany as well as Italy. Table 23 and Figure 15 give a comprehensive view of the sources of economic growth for the G7. The contribution of capital input alone exceeds that of total factor productivity for most nations and most time periods. The contribution of non-IT capital input predominates over IT capital input for most countries and most time periods with Canada in 1989–2001, and the United Kingdom and Japan after 1995 as exceptions. This can be attributed to the unusual weakness in the growth of aggregate demand in these countries. The contribution of labor input varies considerably among the G7 nations with negative contributions after 1995 in Japan, during the 1980s in France, and during the period 1989–1995 in the United Kingdom and Germany. Finally, Table 24 and Figure 16 translate sources of growth into sources of growth in average labor productivity (ALP). Average labor productivity, defined as output per hour worked, must be carefully distinguished from total factor productivity, defined as output per unit of both capital and labor inputs. Output growth is the sum of growth in
Ch. 10:
Accounting for Growth in the Information Age
Figure 15. Sources of economic growth by country.
Figure 16. Sources of labor productivity growth by country.
801
802
D.W. Jorgenson Table 24 Sources of labor productivity growth
Year
U.S.
Canada
1980–1989 1989–1995 1995–2001
3.38 2.43 3.76
3.10 1.39 3.34
1980–1989 1989–1995 1995–2001
1.79 1.02 1.53
1.87 0.20 1.93
1980–1989 1989–1995 1995–2001
1.58 1.40 2.23
1.23 1.19 1.41
1980–1989 1989–1995 1995–2001
0.40 0.44 0.92
0.35 0.48 0.79
1980–1989 1989–1995 1995–2001
0.37 0.34 0.55
0.42 0.16 −0.14
1980–1989 1989–1995 1995–2001
0.30 0.36 0.23
0.40 0.55 0.18
1980–1989 1989–1995 1995–2001
0.23 0.23 0.48
0.14 0.14 0.17
1980–1989 1989–1995 1995–2001
0.29 0.03 0.06
−0.08 −0.14 0.41
UK
France
Germany
Output 2.38 1.99 1.30 2.34 2.34 1.18 Hours 0.82 −0.66 0.11 −1.17 −0.41 −0.71 1.03 0.91 −0.11 Labor productivity 1.87 3.04 1.88 2.79 1.71 3.05 1.71 1.43 1.29 IT capital deepening 0.22 0.19 0.19 0.29 0.20 0.28 0.71 0.39 0.46 Non-IT capital deepening 1.20 2.29 1.20 2.11 1.15 1.33 −0.21 0.25 0.70 Labor quality 0.12 0.24 0.26 0.49 0.61 0.33 0.30 0.19 0.23 Productivity from IT production 0.23 0.29 0.28 0.32 0.29 0.43 0.82 0.56 0.65 Productivity from non-IT production 0.11 0.03 −0.05 −0.43 −0.55 0.69 0.09 0.04 −0.75 2.69 1.62 2.74
Italy
Japan
2.51 1.52 1.90
3.98 2.39 1.89
0.15 −0.57 0.99
0.95 −0.51 −1.14
2.36 2.09 0.92
3.04 2.90 3.03
0.23 0.28 0.45
0.45 0.39 0.85
2.25 1.06 0.61
0.86 1.37 0.96
0.23 0.38 0.35
0.72 0.31 0.37
0.32 0.38 0.68
0.19 0.20 0.43
−0.68 −0.01 −1.17
0.82 0.63 0.42
Note: Percentage. Contributions. Canada data begins in 1981.
hours worked and growth in ALP. Average labor productivity growth depends on the contribution of capital deepening, the contribution of growth in labor quality, and total factor productivity growth. Capital deepening is the contribution of growth in capital input per hour worked and predominates over total factor productivity as a source of ALP growth for the G7 nations. IT capital deepening predominates over non-IT capital deepening in the United States throughout the period 1980–2001 and in Canada after 1989, the United Kingdom,
Ch. 10:
Accounting for Growth in the Information Age
803
and France after 1995. Finally, the contribution of labor quality is positive for all the G7 nations through the period. 3.4. Alternative approaches Edward Denison’s (1967) path-breaking volume, Why Growth Rates Differ, compared differences in growth rates for national income net of capital consumption per capita for the period 1950–1962 with differences of levels in 1960 for eight European countries and the United States. The European countries were characterized by much more rapid growth and a lower level of national income per capita. However, this association did not hold for all comparisons between the individual countries and the United States. Nonetheless, Denison concluded:68 Aside from short-term aberrations Europe should be able to report higher growth rates, at least in national income per person employed, for a long time. Americans should expect this and not be disturbed by it. Maddison (1987, 1991) constructed estimates of aggregate output, input, and total factor productivity growth for France, Germany, Japan, The Netherlands, and the United Kingdom for the period 1870–1987. Maddison (1995) extended estimates for the United States, the United Kingdom, and Japan backward to 1820 and forward to 1992. He defined output as gross of capital consumption throughout the period and constructed constant quality indices of labor input for the period 1913–1984, but not for 1870–1913. Maddison employed capital stock as a measure of the input of capital, ignoring the changes in the composition of capital stock that are such an important source of growth for the G7 nations. This omission is especially critical in assessing the impact of investment in information technology. Finally, he reduced the growth rate of the price index for investment by one percent per year for all countries and all time periods to correct for biases like those identified by Wyckoff (1995). 3.4.1. Comparisons without growth accounts Kuznets (1971) provided elaborate comparisons of growth rates for 14 industrialized countries. Unlike Denison (1967), he did not provide level comparisons. Maddison (1982) filled this lacuna by comparing levels of national product for 16 countries. These comparisons used estimates of purchasing power parities by Irving Kravis, Alan Heston, and Robert Summers (1978).69 Maddison (1995) extended his long-term estimates of the growth of national product and population to 56 countries, covering the period 1820–1992. Maddison (2001)
68 See Denison (1967), especially Chapter 21, “The Sources of Growth and the Contrast between Europe and
the United States”, pp. 296–348. 69 For details see Maddison (1982, pp. 159–168).
804
D.W. Jorgenson
updated these estimates to 1998 in his magisterial volume, The World Economy: A Millennial Perspective. He provided estimates for 134 countries, as well as seven regions of the world – Western Europe, Western Offshoots (Australia, Canada, New Zealand, and the United States), Eastern Europe, Former USSR, Latin America, Asia, and Africa. Purchasing power parities have been updated by successive versions of the Penn World Table. A complete list of these tables through Mark 5 is given by Summers and Heston (1991). The current version of the Penn World Table is available on the Center for International Comparisons website at the University of Pennsylvania (CICUP). This covers 168 countries for the period 1950–2000 and represents one of the most significant achievements in economic measurement of the post-war period.70 3.4.2. Convergence Data presented by Kuznets (1971), Maddison, and successive versions of the Penn World Table have made it possible to reconsider the issue of convergence raised by Denison (1967). Moses Abramovitz (1986) was the first to take up the challenge by analyzing convergence of output per capita among Maddison’s 16 countries. He found that convergence characterized the post-war period, while there was no tendency toward convergence before 1914 and during the inter-war period. Baumol (1986) formalized these results by running a regression of growth rate of GDP per capita over the period 1870–1979 on the 1870 level of GDP per capita.71 In a highly innovative paper on “Crazy explanations for the productivity slowdown” Paul Romer (1987) derived Baumol’s “growth regression” from Solow’s (1970) growth model with a Cobb-Douglas production function. Romer’s empirical contribution was to extend the growth regressions from Maddison’s (1982) 16 advanced countries to the 115 countries in the Penn World Table (Mark 3). Romer’s key finding was an estimate of the elasticity of output with respect to capital close to three-quarters. The share of capital in GNP implied by Solow’s model was less than half as great. Gregory Mankiw, David Romer and David Weil (1992) defended the traditional framework of Kuznets (1971) and Solow (1970). The empirical part of their study is based on data for 98 countries from the Penn World Table (Mark 4). Like Paul Romer (1987), Mankiw, David Romer, and Weil derived a growth regression from the Solow (1970) model; however, they augmented this by allowing for investment in human capital. The results of Mankiw, David Romer and Weil (1992) provided empirical support for the augmented Solow model. There was clear evidence of the convergence predicted by the model; in addition, the estimated elasticity of output with respect to capital was in 70 See Heston, Summers and Aten (2002). The CICUP website is at: http://pwt.econ.upenn.edu/aboutpwt.
html. 71 Baumol’s “growth regression” has spawned a vast literature, recently summarized by Steven Durlauf and
Danny Quah (1999), Ellen McGrattan and James Schmitz (1999), and Islam (2003). Much of this literature is based on data from successive versions of the Penn World Table.
Ch. 10:
Accounting for Growth in the Information Age
805
line with the share of capital in the value of output. The rate of convergence of output per capita was too slow to be consistent with the 1970 version of the Solow model, but supported the augmented version. 3.4.3. Modeling productivity differences Finally, Islam (1995) exploited an important feature of the Penn World Table overlooked in prior studies. This panel data set contains benchmark comparisons of levels of the national product at five-year intervals, beginning in 1960. This made it possible to test an assumption maintained in growth regressions. These regressions had assumed identical levels of productivity for all countries included in the Penn World Table. Substantial differences in levels of total factor productivity among countries have been documented by Denison (1967), by my papers with Christensen and Cummings (1981), Dougherty (1996, 1997), and Yip (2000) and in Section 2 above. By introducing econometric methods for panel data Islam (1995) was able to allow for these differences. He corroborated the finding of Mankiw, David Romer and Weil (1992) that the elasticity of output with respect to capital input coincided with the share of capital in the value of output. In addition, Islam (1995) found that the rate of convergence of output per capita among countries in the Penn World Table substantiated the unaugmented version of the Solow (1970) growth model. In short, “crazy explanations” for the productivity slowdown, like those propounded by Paul Romer (1987, 1994), were unnecessary. Moreover, the model did not require augmentation by endogenous investment in human capital, as proposed by Mankiw, David Romer and Weil (1992). Islam concluded that differences in technology among countries must be included in econometric models of growth rates. This requires econometric techniques for panel data, like those originated by Gary Chamberlain (1982), rather than the regression methods of Baumol, Paul Romer, and Mankiw, David Romer and Weil. Panel data techniques have now superseded regression methods in modeling differences in output per capita. 3.5. Conclusions I conclude that a powerful surge in investment in information technology and equipment after 1995 characterizes all of the G7 economies. This accounts for a large portion of the resurgence in U.S. economic growth, but contributes substantially to economic growth in the remaining G7 economies as well. Another significant source of the G7 growth resurgence after 1995 is a jump in total factor productivity growth in IT-producing industries. For Japan the dramatic upward leap in the impact of IT investment after 1995 was insufficient to overcome downward pressures from deficient growth of aggregate demand. This manifests itself in declining contributions of non-IT capital and labor inputs. Similar downturns are visible in non-IT capital input in France, Germany, and especially the United Kingdom after 1995.
806
D.W. Jorgenson
These findings are based on new data and new methodology for analyzing the sources of economic growth. Internationally harmonized prices for information technology equipment and software are essential for capturing differences among the G7 nations. Constant quality indices of capital and labor inputs are necessary to incorporate the impacts of investments in information technology and human capital. Exploiting the new data and methodology, I have been able to show that investment in tangible assets is the most important source of economic growth in the G7 nations. The contribution of capital input exceeds that of total factor productivity for all countries for all periods. The relative importance of total factor productivity growth is far less than suggested by the traditional methodology of Kuznets (1971) and Solow (1970), which is now obsolete. The conclusion from Islam’s (1995) research is that the Solow (1970) model is appropriate for modeling the endogenous accumulation of tangible assets. It is unnecessary to endogenize human capital accumulation as well. The transition path to balanced growth equilibrium after a change in policies that affects investment in tangible assets requires decades, while the transition after a change affecting investment in human capital requires as much as a century.
4. Economics on internet time The steadily rising importance of information technology has created new research opportunities in all areas of economics. Economic historians, led by Chandler (2000) and Moses Abramovitz and Paul David (1999, 2001),72 have placed the information age in historical context. Abramovitz and David present sources of U.S. economic growth for the nineteenth and twentieth centuries. Their estimates, beginning in 1966, are based on the official productivity statistics published by the Bureau of Labor Statistics (1994). The Solow (1987) Paradox, that we see computers everywhere but in the productivity statistics,73 has been displaced by the economics of the information age. Computers have now left an indelible imprint on the productivity statistics. The remaining issue is whether the breathtaking speed of technological change in semiconductors differentiates this resurgence from previous periods of rapid growth? Capital and labor markets have been severely impacted by information technology. Enormous uncertainty surrounds the relationship between equity valuations and future growth prospects of the American economy.74 One theory attributes rising valuations of equities since the growth acceleration began in 1995 to the accumulation of intangible assets, such as intellectual property and organizational capital. An alternative theory 72 See also: David (1990, 2000) and Gordon (2000). 73 Griliches (1994), Brynjolfsson and Shinkyu Yang (1996), and Triplett (1999) discuss the Solow Paradox. 74 Campbell and Shiller (1998) and Shiller (2000) discuss equity valuations and growth prospects. Michael
Kiley (1999), Brynjolfsson and Hitt (2000), and Robert Hall (2000, 2001), present models of investment with internal costs of adjustment.
Ch. 10:
Accounting for Growth in the Information Age
807
treats the high valuations of technology stocks as a bubble that burst during the year 2000. The behavior of labor markets also poses important puzzles. Widening wage differentials between workers with more and less education has been attributed to computerization of the workplace. A possible explanation could be that high-skilled workers are complementary to IT, while low-skilled workers are substitutable. An alternative explanation is that technical change associated with IT is skill-biased and increases the wages of high-skilled workers relative to low-skilled workers.75 Finally, information technology is altering product markets and business organizations, as attested by the large and growing business literature,76 but a fully satisfactory model of the semiconductor industry remains to be developed.77 Such a model would derive the demand for semiconductors from investment in information technology in response to rapidly falling IT prices. An important objective is to determine the product cycle for successive generations of new semiconductors endogenously. The semiconductor industry and the information technology industries are global in their scope with an elaborate international division of labor.78 This poses important questions about the American growth resurgence. Where is the evidence of a new economy in other leading industrialized countries? I have shown in Section 3 that the most important explanation is the relative paucity of constant quality price indexes for semiconductors and information technology in national accounting systems outside the United States. The stagflation of the 1970s greatly undermined the Keynesian Revolution, leading to a New Classical Counter-revolution led by Lucas (1981) that has transformed macroeconomics. The unanticipated American growth revival of the 1990s has similar potential for altering economic perspectives. In fact, this is already foreshadowed in a steady stream of excellent books on the economics of information technology.79 We are the fortunate beneficiaries of a new agenda for economic research that will refresh our thinking and revitalize our discipline. Acknowledgements The Program on Technology and Economic Policy at Harvard University provided financial support. The Economic and Social Research Institute of the Cabinet Office of 75 Daron Acemoglu (2000) and Katz (2000) survey the literature on labor markets and technological change. 76 See, for example, Andrew Grove (1996) on the market for computers and semiconductors and Clayton
Christensen (1997) on the market for storage devices. 77 Douglas Irwin and Peter Klenow (1994), Flamm (1996, pp. 305–424), and Elhanan Helpman and Manuel
Trajtenberg (1998, pp. 111–119), present models of the semiconductor industry. 78 The role of information technology in U.S. economic growth is discussed by the Economics and Statistics
Administration (2000); comparisons among OECD countries are given by the Organization for Economic Co-operation and Development (2000, 2003). 79 See, for example, Carl Shapiro and Hal Varian (1999), Brynjolfsson and Kahin (2000), and Choi and Whinston (2000).
808
D.W. Jorgenson
the Government of Japan supported the research reported in Section 4 from its program for international collaboration through the Nomura Research Institute. I am greatly indebted to Jon Samuels for excellent research assistance, as well as useful comments. J. Steven Landefeld, Clinton McCully, and David Wasshausen of the Bureau of Economic Analysis provided valuable data on information technology in the U.S. Tom Hale, Mike Harper, Tom Nardone and Larry Rosenblum (BLS), Kurt Kunze (BEA), Eldon Ball (ERS), Mike Dove and Scott Segerman (DMDC) also provided data for the U.S. and helpful advice. I am grateful to John Baldwin and Tarek Harchaoui of Statistics Canada for data on Canada, Kazuyuki Motohashi and Koji Nomura for data on Japan, and Alessandra Colecchia, Marcel Timmer and Bart Van Ark for data on Europe. Colleagues far too numerous to mention have contributed useful suggestions. I am grateful to all of them but retain sole responsibility for any remaining deficiencies.
References Abramovitz, M. (1956). “Resources and output trends in the united states since 1870”. American Economic Review 46 (1), 5–23. Abramovitz, M. (1986). “Catching up, forging ahead, and falling behind”. Journal of Economic History 46 (2), 385–406. Abramovitz, M., David, P. (1999). “American macroeconomic growth in the era of knowledge-based progress: The long-run perspective”. In: Gallman, R.E., Engerman, S.I. (Eds.), Cambridge Economic History of the United States. Cambridge University Press, Cambridge, MA, pp. 1–92. Abramovitz, M., David, P. (2001). “Two centuries of American macroeconomic growth from exploitation of resource abundance to knowledge-driven development”. Stanford, Stanford Institute for Economic Policy Research, Policy Paper No. 01-005. August. Acemoglu, D. (2000). “Technical change, inequality, and the labor market”. Journal of Economic Literature 40 (1), 7–72. Aizcorbe, A., Oliner, S.D., Sichel, D.E. (2003). “Trends in semiconductor prices: Breaks and explanations”. Board of Governors of the Federal Reserve System, Washington. July. Baily, M.N. (2002). “The new economy: Post mortem or second wind?”. Journal of Economic Perspectives 16 (1), 3–22. Baily, M.N., Gordon, R.J. (1988). “The productivity slowdown, measurement issues, and the explosion of computer power”. Brookings Papers on Economic Activity 2, 347–420. Baldwin, J.R., Harchaoui, T.M. (2002). Productivity Growth in Canada – 2002. Ottawa, Statistics Canada. Baumol, W.J. (1986). “Productivity growth, convergence, and welfare”. American Economic Review 76 (5), 1072–1085. Becker, G.S. (1993a). Human Capital, third ed. University of Chicago Press, Chicago. First ed., 1964; second ed., 1975. Becker, G.S. (1993b). “Nobel lecture: The economic way of looking at behavior”. Journal of Political Economy 101 (3), 385–409. Blades, D. (2001). Measuring Capital: A Manual on the Measurement of Capital Stocks, Consumption of Fixed Capital, and Capital Services. Organization for Economic Co-operation and Development, Paris. Bosworth, B.P., Triplett, J. (2000). “What’s new about the new economy? IT, growth and productivity”. The Brookings Institution, Washington. October 20. Brynjolfsson, E., Hitt, L.M. (2000). “Beyond computation: Information technology, organizational transformation and business performance”. Journal of Economic Perspectives 14 (4), 23–48. Fall. Brynjolfsson, E., Kahin, B. (Eds.) (2000). Understanding the Digital Economy. The MIT Press, Cambridge, MA.
Ch. 10:
Accounting for Growth in the Information Age
809
Brynjolfsson, E., Yang, S. (1996). “Information technology and productivity: A review of the literature”. Advances in Computers 43 (1), 179–214. Bureau of Economic Analysis (1986). “Improved deflation of purchase of computers”. Survey of Current Business 66 (3), 7–9. Bureau of Economic Analysis (1995). “Preview of the comprehensive revision of the national income and product accounts: Recognition of government investment and incorporation of a new methodology for calculating depreciation”. Survey of Current Business 75 (9), 33–41. Bureau of Economic Analysis (1999). Fixed Reproducible Tangible Wealth in the United States, 1925–94. U.S. Department of Commerce, Washington. Bureau of Labor Statistics (1983). Trends in Multifactor Productivity, 1948–1981. U.S. Government Printing Office, Washington. Bureau of Labor Statistics (1994). “Multifactor productivity measures, 1991 and 1992.” News Release USDL 94-327. July 11. Campbell, J.Y., Shiller, R.J. (1998). “Valuation ratios and the long-run stock market outlook”. Journal of Portfolio Management 24 (2), 11–26. Chamberlain, G. (1982). “Multivariate regression models for panel data”. Journal of Econometrics 18 (1), 5–46. Chandler, A.D. Jr. (2000). “The information age in historical perspective”. In: Chandler, A.D., Cortada, J.W. (Eds.), A Nation Transformed by Information: How Information Has Shaped the United States from Colonial Times to the Present. Oxford University Press, New York, pp. 3–38. Choi, S.-Y., Whinston, A.B. (2000). The Internet Economy: Technology and Practice. SmartEcon Publishing, Austin. Chow, G.C. (1967). “Technological change and the demand for computers”. American Economic Review 57 (5), 1117–1130. Christensen, C.M. (1997). The Innovator’s Dilemma. Harvard Business School Press, Boston. Christensen, L.R., Cummings, D., Jorgenson, D.W. (1980). “Economic growth, 1947–1973: An international comparison”. In: Kendrick, J.W., Vaccara, B. (Eds.), New Developments in Productivity Analysis. University of Chicago Press, Chicago, pp. 595–698. Christensen, L.R., Cummings, D., Jorgenson, D.W. (1981). “Relative productivity levels, 1947–1973”. European Economic Review 16 (1), 61–94. Colecchia, A., Schreyer, P. (2002). “ICT Investment and economic growth in the 1990s: Is the United States a unique case? A comparative study of nine OECD countries”. Review of Economic Dynamics 5 (2), 408–442. Congressional Budget Office (2002). The Budget and Economic Outlook: An Update. U.S. Government Printing Office, Washington, DC. Corrado, C. (2003). “Industrial production and capacity utilization: The 2002 historical and annual revision”. Federal Reserve Bulletin, April, pp. 151–176. David, P.A. (1990). “The dynamo and the computer: An historical perspective on the productivity paradox”. American Economic Review 80 (2), 355–361. David, P.A. (2000). “Understanding digital technology’s evolution and the path of measured productivity growth: present and future in the mirror of the past”. In: Brynjolfsson, E., Kahin, B. (Eds.), Understanding the Digital Economy. The MIT Press, Cambridge, MA, pp. 49–98. Dean, E.R., Harper, M.J. (2001). “The BLS productivity measurement program”. In: Hulten, C.R., Dean, E.R., Harper, M.J. (Eds.), New Developments in Productivity Analysis, pp. 55–84. Denison, E.F. (1962). The Sources of Economic Growth in the United States and the Alternatives before Us. Committee on Economic Development, New York. Denison, E.F. (1967). Why Growth Rates Differ. The Brookings Institution, Washington. Denison, E.F. (1974). Accounting for United States Economic Growth, 1929 to 1969. The Brookings Institution, Washington, DC. Denison, E.F. (1989). Estimates of Productivity Change by Industry. Brookings Institution, Washington. Dertouzos, M., Solow, R.M., Lester, R.K. (1989). Made in American: Regaining the Productive Edge. The MIT Press, Cambridge, MA.
810
D.W. Jorgenson
Diewert, W.E. (1976). “Exact and superlative index numbers”. Journal of Econometrics 4 (2), 115–146. Diewert, W.E. (1980). “Aggregation problems in the measurement of capital”. In: Usher, D. (Ed.), The Measurement of Capital. University of Chicago Press, Chicago, pp. 433–528. Diewert, W.E., Lawrence, D.A. (2000). “Progress in measuring the price and quantity of capital”. In: Lau (Ed.), Econometrics and the Cost of Capital. The MIT Press, Cambridge, MA, pp. 273–326. Domar, E. (1946). “Capital expansion, rate of growth and employment”. Econometrica 14 (2), 137–147. Domar, E. (1961). “On the measurement of technological change”. Economic Journal 71 (284), 709–729. Doms, M. (2004). “Communications equipment: What has happened to prices?”. In: Corrado, C., Haltiwanger, J., Hulten, C. (Eds.), Measurement of Capital in the New Economy. University of Chicago Press, Chicago. Dougherty, C., Jorgenson, D.W. (1996). “International comparisons of the sources of economic growth”. American Economic Review 86 (2), 25–29. Dougherty, C., Jorgenson, D.W. (1997). “There is no silver bullet: Investment and growth in the G7”. National Institute Economic Review (162), 57–74. Douglas, P.H. (1948). “Are there laws of production?”. American Economic Review 38 (1), 1–41. Dulberger, E.R. (1989). “The application of a hedonic model to a quality-adjusted price index for computer processors”. In: Jorgenson, D.W., Landau, R. (Eds.), Technology and Capital Formation. The MIT Press, Cambridge, MA, pp. 37–76. Dulberger, E.R. (1993). “Sources of decline in computer processors: Selected electronic components”. In: Foss, M.F., Manser, M.E., Young, A.H. (Eds.), Price Measurements and Their Uses. University of Chicago Press, Chicago, pp. 103–124. Durlauf, S.N., Quah, D.T. (1999). “The new empirics of economic growth”. In: Taylor, J., Woodford, M. (Eds.), Handbook of Macroeconomics, vol. 1A. North-Holland, Amsterdam, pp. 235–310. Easterly, W. (2001). The Elusive Quest for Growth. The MIT Press, Cambridge, MA. Economics and Statistics Administration (2000). Digital Economy 2000. U.S. Department of Commerce, Washington, DC. Fisher, I. (1922). The Making of Index Numbers. Houghton-Mifflin, Boston. Flamm, K. (1989). “Technological advance and costs: Computers versus communications”. In: Crandall, R.C., Flamm, K. (Eds.), Changing the Rules: Technological Change, International Competition, and Regulation in Communications. Brookings Institution Press, Washington, DC, pp. 13–61. Flamm, K. (1996). Mismanaged Trade? Strategic Policy and the Semiconductor Industry. Brookings Institution Press, Washington, DC. Fraumeni, B.M. (1997). “The measurement of depreciation in the U.S. national income and product accounts”. Survey of Current Business 77 (7), 7–23. Goldsmith, R. (1962). The National Wealth of the United States in the Postwar Period. National Bureau of Economic Research, New York. Gollop, F.M. (2000). “The cost of capital and the measurement of productivity”. In: Lau, L. (Ed.), Econometrics and the Cost of Capital. The MIT Press, Cambridge, MA, pp. 85–110. Gordon, R.J. (1989). “The postwar evolution of computer prices”. In: Jorgenson, D.W., Landau, R. (Eds.), Technology and Capital Formation. The MIT Press, Cambridge, MA, pp. 77–126. Gordon, R.J. (1990). The Measurement of Durable Goods Prices. University of Chicago Press, Chicago. Gordon, R.J. (1998). “Foundations of the Goldilocks economy: Supply shocks and the time-varying NAIRU”. Brookings Papers on Economic Activity 2, 297–333. Gordon, R.J. (2000). “Does the ‘New Economy’ measure up to the great inventions of the past”. Journal of Economic Perspectives 14 (4), 49–74. Fall. Greenspan, A. (2000). “Challenges for monetary policy-makers”. Board of Governors, Federal Reserve System, Washington (October 19). Greenwood, J., Hercowitz, Z., Krusell, P. (1997). “Long-run implications of investment-specific technological change”. American Economic Review 87 (3), 342–362. Greenwood, J., Hercowitz, Z., Krusell, P. (2000). “The role of investment-specific technological change in the business cycle”. European Economic Review 44 (1), 91–115.
Ch. 10:
Accounting for Growth in the Information Age
811
Greenwood, J., Jovanovic, B. (2001). “Accounting for growth”, In: Hulten, Charles R., Dean, Edvin R., Harper, Michael J. (Eds.), New Developments in Productivity Analysis, University of Chicago Press, Chicago, pp. 179–222. Griliches, Z. (1960). “Measuring inputs in agriculture: A critical survey”. Journal of Farm Economics 40 (5), 1398–1427. Griliches, Z. (1994). “Productivity, R&D, and the data constraint”. American Economic Review 94 (2), 1–23. Griliches, Z. (2000). R&D, Education, and Productivity. Harvard University Press, Cambridge, MA. Grimm, B.T. (1997). “Quality adjusted price indexes for digital telephone switches”. Bureau of Economic Analysis, Washington. May 20. Grimm, B.T. (1998). “Price indexes for selected semiconductors: 1974–1996”. Survey of Current Business 78 (2), 8–24. Grove, A.S. (1996). Only the Paranoid Survive: How to Exploit the Crisis Points that Challenge Every Company. Doubleday, New York. Gullickson, W., Harper, M.J. (1999). “Possible measurement bias in aggregate productivity growth”. Monthly Labor Review 122 (2), 47–67. Hall, R.E. (1988). “The relation between price and marginal cost in U.S. industry”. Journal of Political Economy 96 (5), 921–947. Hall, R.E. (1990a). “Invariance properties of Solow’s productivity residual”. In: Diamond, P. (Ed.), Growth/Productivity/Employment. The MIT Press, Cambridge, MA. Hall, R.E. (2000). “e-Capital: The link between the stock market and the labor market in the 1990s”. Brookings Papers on Economic Activity 2, 73–118. Hall, R.E. (2001). “The stock market and capital accumulation”. American Economic Review 91, 1185–1202. Hall, R.E. (2002). Digital Dealing: How E-Markets Are Transforming the Economy. Norton, New York. Harberger, A.C. (1998). “A vision of the growth process”. American Economic Review 88 (1), 1–32. Harrod, R. (1939). “An essay in dynamic theory”. Economic Journal 49 (194), 14–33. Hayashi, F. (2000). “The cost of capital, Q, and the theory of investment demand”. In: Lau, L. (Ed.), Econometrics and the Cost of Capital. The MIT Press, Cambridge MA, pp. 55–84. Hecht, J. (1999). City of Light. Oxford University Press, New York. Helpman, E., Trajtenberg, M. (1998). “Diffusion of general purpose technologies”. In: Helpman, E. (Ed.), General Purpose Technologies and Economic Growth. The MIT Press, Cambridge, MA, pp. 85–120. Hercowitz, Z. (1998). “The ‘Embodiment’ controversy: A review essay”. Journal of Monetary Economics 41 (1), 217–224. Heston, A., Summers, R., Aten, B. (2002). “Penn World Table Version 6.1”. Center for International Comparisons and the University of Pennsylvania (CICUP). October. Hulten, C.R. (1990). “The measurement of capital”. In: Berndt, E.R., Triplett, J. (Eds.), Fifty Years of Economic Measurement. University of Chicago Press, Chicago, pp. 119–152. Hulten, C.R. (1992). “Growth accounting when technical change is embodied in capital”. American Economic Review 82 (4), 964–980. Hulten, C.R. (2001). “Total factor productivity: A short biography”. In: Hulten, C.R., Dean, E.R., Harper, M.J. (Eds.), New Developments in Productivity Analysis. University of Chicago Press, Chicago, pp. 1–47. International Technology Roadmap for Semiconductors, 2004 Update (December 2004). Sematech Corporation, Austin. Irwin, D.A., Klenow, P.J. (1994). “Learning-by-doing spillovers in the semiconductor industry”. Journal of Political Economy 102 (6), 1200–1227. Islam, N. (1995). “Growth empirics”. Quarterly Journal of Economics 110 (4), 1127–1170. Islam, N. (2003). “What have we learned from the convergence debate?”. Journal of Economic Surveys 17 (3), 309–362. Jorgenson, D.W. (1963). “Capital theory and investment behavior”. American Economic Review 53 (2), 247– 259. Jorgenson, D.W. (1966). “The embodiment hypothesis”. Journal of Political Economy 74 (1), 1–17. Jorgenson, D.W. (1973). “The economic theory of replacement and depreciation”. In: Sellekaerts, W. (Ed.), Econometrics and Economic Theory. Macmillan, New York, pp. 189–221.
812
D.W. Jorgenson
Jorgenson, D.W. (1980). “Accounting for capital”. In: von Furstenberg, G.M. (Ed.), Capital, Efficiency and Growth. Ballinger, Cambridge, MA, pp. 251–319. Jorgenson, D.W. (1989). “Capital as a factor of production”. In: Jorgenson, D.W., Landau, R. (Eds.), Technology and Capital Formation, pp. 1–36. Jorgenson, D.W. (1990). “Productivity and economic growth”. In: Berndt, E.R., Triplett, J. (Eds.), Fifty Years of Economic Measurement. University of Chicago Press, Chicago, pp. 19–118. Jorgenson, D.W. (1996). “Empirical studies of depreciation”. Economic Inquiry 34 (1), 24–42. Jorgenson, D.W. (2001). “Information technology and the U.S. economy”. American Economic Review 91 (1), 1–32. Jorgenson, D.W., Gollop, F.M., Fraumeni, B.M. (1987). Productivity and U.S. Economic Growth. Harvard University Press, Cambridge, MA. Jorgenson, D.W., Griliches, Z. (1967). “The explanation of productivity change”. Review of Economic Studies 34 (99), 249–280. Jorgenson, D.W., Ho, M.S., Stiroh, K.J. (2004). “Growth of U.S. industries and investments in information technology and higher education”. In: Corrado, C., Hulten, C., Sichel, D. (Eds.), Measuring Capital in a New Economy. University of Chicago Press, Chicago. Jorgenson, D.W., Motohashi, K. (2003). “Economic growth of Japan and the U.S. in the information age”. Research Institute of Economy, Trade, and Industry, Tokyo. July. Jorgenson, D.W., Stiroh, K.J. (1995). “Computers and growth”. Economics of Innovation and New Technology 3 (3–4), 295–316. Jorgenson, D.W., Stiroh, K.J. (1999). “Information technology and growth”. American Economic Review 89 (2), 109–115. Jorgenson, D.W., Stiroh, K.J. (2000a). “U.S. economic growth at the industry level”. American Economic Review 90 (2), 161–167. Jorgenson, D.W., Stiroh, K.J. (2000b). “Raising the speed limit: U.S. economic growth in the information age”. Brookings Papers on Economic Activity 1, 125–211. Jorgenson, D.W., Yip, E. (2000). “Whatever happened to productivity growth?”. In: Hulten, C.R., Dean, E.R., Harper, M.J. (Eds.), New Developments in Productivity Analysis. University of Chicago Press, Chicago, pp. 509–540. Jorgenson, D.W., Yun, K.-Y. (2001). Lifting the Burden: Tax Reform, the Cost of Capital, and U.S. Economic Growth. The MIT Press, Cambridge, MA. Katz, L.F. (2000). “Technological change, computerization, and the wage structure”. In: Brynjolfsson, E., Kahin, B. (Eds.), Understanding the Digital Economy. The MIT Press, Cambridge, MA, pp. 217–244. Katz, L.F., Krueger, A. (1999). “The high-pressure U.S. labor market of the 1990s”. Brookings Papers on Economic Activity 1, 1–87. Kendrick, J.W. (1956). “Productivity trends: Capital and labor”. Review of Economics and Statistics 38 (3), 248–257. Kendrick, J.W. (1961). Productivity Trends in the United States. Princeton University Press, Princeton. Kendrick, J.W. (1973). Postwar Productivity Trends in the United States. National Bureau of Economic Research, New York. Kendrick, J.W., Grossman, E. (1980). Productivity in the United States: Trends and Cycles. Johns Hopkins Press, Baltimore, MD. Kiley, M.T. (1999). “Computers and growth with costs of adjustment: Will the future look like the past?”. Board of Governors of the Federal Reserve System, Washington, DC. July. King, R.G., Rebelo, S. (1999). “Resuscitating real business cycles”. In: Taylor, J., Woodford, M. (Eds.), Handbook of Microeconomics, vol. 1B. North-Holland, Amsterdam, pp. 927–1008. Konus, A., Byushgens, S.S. (1926). “On the problem of the purchasing power of money”. Economic Bulletin of the Conjuncture Institute, 151–172. Supplement. Kravis, I.B., Heston, A., Summers, R. (1978). International Comparisons of Real Product and Purchasing Power. Johns Hopkins University Press, Baltimore, MD. Kuznets, S. (1961). Capital in the American Economy. Princeton University Press, Princeton.
Ch. 10:
Accounting for Growth in the Information Age
813
Kuznets, S. (1971). Economic Growth of Nations. Harvard University Press, Cambridge, MA. Landefeld, J.S., Parker, R.P. (1997). “BEA’s chain indexes, time series, and measures of long-term growth”. Survey of Current Business 77 (5), 58–68. Lau, L. (Ed.) (2000). Econometrics and the Cost of Capital. The MIT Press, Cambridge, MA. Lindbeck, A. (Ed.) (1992). Nobel Lectures in Economic Sciences, 1969–1980. World Scientific, River Edge, NJ. Lucas, R.E. Jr. (1967). “Adjustment costs and the theory of supply”. Journal of Political Economy 75 (4), 321–334. Lucas, R.E. (1981). Studies in Business-Cycle Theory. The MIT Press, Cambridge, MA. Maddison, A. (1982). Phases of Capitalist Development. Oxford University Press, Oxford. Maddison, A. (1987). “Growth and slowdown in advanced capitalist economies: Techniques of quantitative assessment”. Journal of Economic Literature 25 (2), 649–698. Maddison, A. (1991). Dynamic Forces in Capitalist Development. Oxford University Press, Oxford. Maddison, A. (1995). Monitoring the World Economy. Organization for Economic Co-operation and Development, Paris. Maddison, A. (2001). The World Economy: A Millennial Perspective. Organization for Economic Cooperation and Development, Paris. Maler, K.-G. (Ed.) (1992). Nobel Lectures in Economic Sciences, 1981–1990. World Scientific, River Edge, NJ. Mankiw, N.G., Romer, D., Weil, D. (1992). “A contribution to the empirics of economic growth”. Quarterly Journal of Economics 107 (2), 407–437. McGrattan, E., Schmitz, J. (1999). In: Taylor, J., Woodford, M. (Eds.), Handbook of Macroeconomics, vol. 1A. North-Holland, Amsterdam, pp. 669–737. Moore, G.E. (1965). “Cramming more components onto integrated circuits”. Electronics 38 (8), 114–117. Moore, G.E. (1996). “Intel – Memories and the Microprocessor”. Daedalus 125 (2), 55–80. Moore, G.E. (1997). “An update on Moore’s Law”. Intel Corporation, Santa Clara, CA. September 30. Moore, G.E. (2003). “No exponential is forever . . . but we can delay forever”. Intel Corporation, Santa Clara, CA. February 10. Moulton, B.R. (2000). “Improved estimates of the national income and product accounts for 1929–1999: Results of the comprehensive revision”. Survey of Current Business 80 (4), 11–17, 36–145. Oliner, S.D., Sichel, D.E. (1994). “Computers and output growth revisited: how big is the puzzle?”. Brookings Papers on Economic Activity 2, 273–317. Oliner, S.D., Sichel, D.E. (2000). “The resurgence of growth in the late 1990s: Is information technology the story?”. Journal of Economic Perspectives 14 (4), 3–22. Organization for Economic Co-operation and Development (2000). A New Economy?. Organization for Economic Co-operation and Development, Paris. Organization for Economic Co-operation and Development (2003). ICT and Economic Growth. Organization for Economic Co-operation and Development, Paris. Parker, R.P. Grimm, B.T. (2000). “Recognition of business and government expenditures on software as investment: Methodology and quantitative impacts, 1959–98”. Bureau of Economic Analysis, Washington. November 14. Petzold, C. (1999). Code: The Hidden Language of Computer Hardware and Software. Microsoft Press, Redmond. Rashad, R. (2000). “The future – it isn’t what it used to be”. Microsoft Research, Seattle. May 3. Rees, A. (Ed.) (1979). Measurement and Interpretation of Productivity. National Academy Press, Washington. Romer, P. (1987). “Crazy explanations for the productivity slowdown”. In: Fischer, S. (Ed.), NBER Macroeconomics Annual. The MIT Press, Cambridge, MA, pp. 163–201. Romer, P. (1994). “The origins of endogenous growth”. Journal of Economic Perspectives 8 (1), 3–20. Ruttan, V.W. (2001). Technology, Growth, and Development. Oxford University Press, New York. pp. 316– 367. Chapter “The computer and semiconductor industries”.
814
D.W. Jorgenson
Schreyer, P. (2000). “The contribution of information and communication technology to output growth: A study of the G7 countries”. Organization for Economic Co-operation and Development, Paris. May 23. Schreyer, P. (2001). OECD Productivity Manual: A Guide to the Measurement of Industry-Level and Aggregate Productivity Growth. Organization for Economic Co-operation and Development, Paris. Shapiro, C., Varian, H.R. (1999). Information Rules. Harvard Business School Press, Boston. Shiller, R. (2000). Irrational Exuberance. Princeton University Press, Princeton. Solow, R.M. (1956). “A contribution to the theory of economic growth”. Quarterly Journal of Economics 70 (1), 65–94. Solow, R.M. (1957). “Technical change and the aggregate production function”. Review of Economics and Statistics 39 (3), 312–320. Solow, R.M. (1960). “Investment and technical progress”. In: Arrow, K.J., Karlin, S., Suppes, P. (Eds.), Mathematical Methods in the Social Sciences, 1959. Stanford University Press, Stanford, pp. 89–104. Solow, R.M. (1970). Growth Theory: An Exposition. Oxford University Press, New York. Solow, R. (1987). “We’d better watch out”. New York Review of Books, July 12. Solow, R.M. (1988). “Growth theory and after”. American Economic Review 78 (3), 307–317. Solow, R.M. (1999). “Neoclassical growth theory”. In: Taylor, J., Woodford, M. (Eds.), Handbook of Macroeconomics, vol. 1A. North-Holland, Amsterdam, pp. 637–668. Solow, R. (2001). “After technical progress and the aggregate production function”. In: Hulten, C.R., Dean, E.R., Harper, M.J. (Eds.), New Developments in Productivity Analysis. University of Chicago Press, Chicago, pp. 173–178. Stigler, G.J. (1947). Trends in Output and Employment. National Bureau of Economic Research, New York. Stiroh, K.J. (2002). “Information technology and the U.S. productivity revival: What do the industry data say?”. American Economic Review 92 (5), 1559–1576. Summers, R., Heston, A. (1991). “The Penn World Table (Mark 5): An expanded set of international comparisons, 1950–1988”. Quarterly Journal of Economics 106 (2), 327–368. Tinbergen, J. (1959). “On the theory of trend movements”. In: Tinbergen, J. (Ed.), Selected Papers. NorthHolland, Amsterdam, pp. 182–221. (Translated from “Zur Theorie der langfristigen Wirtschaftsentwicklung”, Weltwirtschaftliches Archiv 55 (1) (1942), 511–549.) Tobin, J. (1969). “A general equilibrium approach to monetary theory”. Journal of Money, Credit, and Banking 1 (1), 15–29. Triplett, J.E. (1986). “The economic interpretation of hedonic methods”. Survey of Current Business 66 (1), 36–40. Triplett, J.E. (1989). “Price and technological change in a capital good: Survey of research on computers”. In: Jorgenson, D.W., Landau, R. (Eds.), Technology and Capital Formation. The MIT Press, Cambridge, MA, pp. 127–213. Triplett, J.E. (1996). “High-tech industry productivity and hedonic price indices”. In: Industry Productivity. Organization for Economic Co-operation and Development, Paris, pp. 119–142. Triplett, J.E. (1999). “The Solow productivity paradox: What do computers do to productivity?”. Canadian Journal of Economics 32 (2), 309–334. Triplett, J.E. (2003). Handbook on the Quality Adjustment of Price Indices for ICT Products. Organization for Economic Co-Operation and Development, Paris. United Nations (1993). System of National Accounts 1993. United Nations, New York. Van Ark, B., Melka, J., Mulder, N., Timmer, M., Ypma, G. (2002). ICT Investment and Growth Accounts for the European Union, 1980–2000. European Commission, Brussels. Whelan, K. (2002). “Computers, obsolescence, and productivity”. Review of Economics and Statistics 84 (3), 445–462. Wolfe, T. (2000). “Two men who went West”. In: Hooking Up. Farrar, Straus, and Giroux, New York, pp. 17– 65. Wyckoff, A.W. (1995). “The impact of computer prices on international comparisons of productivity”. Economics of Innovation and New Technology 3 (34), 277–293. Young, A. (1989). “BEA’s measurement of computer output”. Survey of Current Business 69 (7), 108–115.
Ch. 10:
Accounting for Growth in the Information Age
815
Further reading Berndt, E.R., Triplett, J. (Eds.) (1990). Fifty Years of Economic Measurement. University of Chicago Press, Chicago. Cole, R., Chen, Y.C., Barquin-Stolleman, J.A., Dulberger, E.R., Helvacian, N., Hodge, J.H. (1986). “Qualityadjusted price indexes for computer processors and selected peripheral equipment”. Survey of Current Business 66 (1), 41–50. Corrado, C., Haltiwanger, J., Hulten, C. (Eds.) (2004). Measurement of Capital in the New Economy. University of Chicago Press, Chicago. Herman, S.W. (2000). “Fixed assets and consumer durable goods for 1925–1999”. Survey of Current Business 80 (9), 19–30. Hulten, C.R., Dean, E.R., Harper, M.J. (Eds.) (2001). New Developments in Productivity Analysis. University of Chicago Press, Chicago. Jorgenson, D., Landau, R. (Eds.) (1989). Technology and Capital Formation. The MIT Press, Cambridge, MA. Kendrick, J., Vaccara, B. (Eds.) (1980). New Developments in Productivity Measurement and Analysis. University of Chicago Press, Chicago. Moss, M. (Ed.) (1973). The Measurement of Economic and Social Performance. Columbia University Press, New York. Summers, R., Heston, A. (1984). “Improved international comparisons of real product and its composition: 1950–1980”. Review of Income and Wealth 30 (1), 1–25. (Mark 3). Summers, R., Heston, A. (1988). “A new set of international comparisons of real product and price levels: Estimates for 130 countries, 1950–1985”. Review of Income and Wealth 34 (1), 19–26. (Mark 4). Taylor, J., Woodford, M. (Eds.) (1999). Handbook of Macroeconomics, 3 vols. North-Holland, Amsterdam. Usher, D. (Ed.) (1980). The Measurement of Capital. University of Chicago Press, Chicago.
Chapter 11
EXTERNALITIES AND GROWTH PETER J. KLENOW Stanford University and NBER e-mail:
[email protected] ANDRÉS RODRÍGUEZ-CLARE Pennsylvania State University e-mail:
[email protected]
Contents Abstract Keywords 1. Introduction 2. A brief guide to externalities in growth models 2.1. 2.2. 2.3. 2.4.
Models with knowledge externalities Models with knowledge externalities and new-good externalities Models with new-good externalities Models with no externalities
3. Cross-country evidence 3.1. 3.2. 3.3. 3.4. 3.5.
The world-wide growth slowdown Beta convergence in the OECD Low persistence of growth rate differences Investment rates and growth vs. levels R&D and TFP
4. Models with common growth driven by international knowledge spillovers 4.1. 4.2. 4.3. 4.4. 4.5. 4.6.
R&D investment and relative productivity Modeling growth in the world technology frontier Determinants of R&D investment Calibration The benefits of engagement Discussion of main results
5. Conclusion Acknowledgements Appendix A Comparative statics
References Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01011-7
818 818 819 819 820 821 822 823 825 825 827 829 831 833 835 836 839 843 845 854 856 856 857 857 858 859
818
P.J. Klenow and A. Rodríguez-Clare
Abstract Externalities play a central role in most theories of economic growth. We argue that international externalities, in particular, are essential for explaining a number of empirical regularities about growth and development. Foremost among these is that many countries appear to share a common long run growth rate despite persistently different rates of investment in physical capital, human capital, and research. With this motivation, we construct a hybrid of some prominent growth models that have international knowledge externalities. When calibrated, the hybrid model does a surprisingly good job of generating realistic dispersion of income levels with modest barriers to technology adoption. Human capital and physical capital contribute to income differences both directly (as usual), and indirectly by boosting resources devoted to technology adoption. The model implies that most of income above subsistence is made possible by international diffusion of knowledge.
Keywords externalities, spillovers, growth, technology, diffusion JEL classification: O11, O33, O40
Ch. 11:
Externalities and Growth
819
If ideas are the engine of growth and if an excess of social over private returns is an essential feature of the production of ideas, then we want to go out of our way to introduce external effects into growth theory, not try to do without them. Robert E. Lucas (2002, p. 6)
1. Introduction A number of facts suggest that international knowledge externalities are critical for understanding growth and development. The growth slowdown that began in the early 1970s was world-wide, not an OECD-only phenomenon. Countries with high investment rates exhibit higher income levels more than higher growth rates. Country growth rate differences are not very persistent from decade to decade, whereas differences in country incomes and investment rates are highly persistent. These patterns hold for investment rates in physical, human, and research capital. Together, they suggest that investment rates affect country transitional growth rates and long run relative incomes rather than long run growth rates. They also suggest countries are subject to the same long run growth rate. We argue that this represents evidence of very large international spillovers at the heart of the long run growth process. We organize this chapter as follows. In Section 2 we describe two broad types of externalities and the growth models that do (and do not) feature them. Section 3 presents cross-country evidence that, we argue, is very hard to reconcile with the models that have no international externalities. Section 4 calibrates a model of growth with international externalities in the form of technology diffusion. The implied externalities are huge. Section 5 concludes and points out directions for future research.
2. A brief guide to externalities in growth models In this section we briefly discuss the role that externalities play in prominent theories of economic growth. One class of growth theories features externalities in the accumulation of knowledge possessed by firms (organizational capital) or by workers (human capital). Another class of growth models features externalities from the introduction of new goods, in the form of surplus to consumers and/or firms. Other theories combine knowledge externalities and new good externalities. Finally, some important growth theories include no externalities at all. Table 1 provides examples of growth models categorized in these four ways. At the end of this section, we will dwell a little on the predictions of no-externalities models in order to motivate the evidence we describe in the next section. The evidence in the next section will suggest that models with no externalities cannot explain a number of empirical patterns.
820
P.J. Klenow and A. Rodríguez-Clare Table 1 Some growth models by type of externality
Knowledge externalities
No knowledge externalities
New good externalities
No new good externalities
Stokey (1988, 1991) Romer (1990) Aghion and Howitt (1992) Eaton and Kortum (1996) Howitt (1999, 2000) Rivera-Batiz and Romer (1991) Romer (1994) Kortum (1997)
Romer (1986) Lucas (1988, 2004) Tamura (1991) Parente and Prescott (1994) Jones and Manuelli (1990) Rebelo (1991) Acemoglu and Ventura (2002)
2.1. Models with knowledge externalities Romer (1986) modeled endogenous growth due to knowledge externalities: a given firm is more productive the higher the average knowledge stock of other firms. As an example, consider a set of atomistic firms, each with knowledge capital k, benefiting from the average stock of knowledge capital in the economy K in their production of output y: yit = Akitα Kt1−α ,
0 < α < 1.
(2.1)
Romer showed that, under certain conditions, constant returns to economy-wide knowledge, as in this example, can generate endogenous growth. The external effects are, of course, critical for long-run growth given the diminishing returns to private knowledge capital. Romer was agnostic as to whether the knowledge capital should be thought of as disembodied (knowledge in books) or embodied (physical capital and/or human capital). Lucas (1988) was more specific, stressing the importance of human capital. Lucas sketched two models, one with human capital accumulated off-the-job and another with human capital accumulated on-the-job (i.e., learning by doing). Both models featured externalities. In the model with human capital accumulated off-the-job, Lucas posited γ
yit = Akitα [uit hit nit ]1−α Ht ,
with γ > 0,
hit+1 = hit + Bhit [1 − uit ]
with 0 < uit < 1.
and
(2.2) (2.3)
Here u is the fraction of time spent working, and 1 − u is the fraction of time spent accumulating human capital; h is an individual worker’s human capital, and H is economy-wide average human capital; k and n are physical capital and number of workers at a given firm. Because human capital accumulation is linear in the level of human capital, human capital is an engine of growth in this model. This is true with or without the externalities; across-dynasty externalities are not necessary for growth. As Lucas discusses, however, within-dynasty human capital spillovers are implicit if one imagines (2.3) as successive generations of finite-lived individuals within a dynasty.
Ch. 11:
Externalities and Growth
821
Within-dynasty externalities, however, would not have the same normative implications as across-dynasty externalities, namely underinvestment in human capital. Lucas (1988) did not argue that across-dynasty externalities were needed to fit particular facts. But he later observed that such across-household externalities could help explain why we see “immigration at maximal allowable rates and beyond from poor countries to wealthy ones” [Lucas (1990, p. 93)]. Tamura (1991) analyzed a human capital externality in the production of human capital itself. This formulation conformed better to the intuition that individuals learn from the knowledge of others. Tamura specified yit = Akitα [uit hit nit ]1−α , β 1−β hit+1 = hit + B hit [1 − uit ] Ht .
(2.4) (2.5)
Because H represents economy-wide average human capital, β < 1 implies that learning externalities are essential for sustaining growth in Tamura’s setup. If applied to each country, this model would suggest that immigrants from poor to rich countries should enjoy fast wage growth after they migrate, as they learn from being around higher average human capital in richer countries. Lucas (2004) used such learning externalities within cities as an ingredient of a model of urbanization and development. Models not always thought of as having knowledge externalities are Mankiw, Romer and Weil’s (1992) augmented Solow model and the original Solow (1956) neoclassical growth model. In Solow’s model all firms within the economy enjoy the same level of TFP. This common level of TFP reflects technology accessible to all. The Solow model therefore does feature disembodied knowledge externalities across firms within an economy. In Mankiw et al.’s extension, knowledge externalities flow across countries as well as across firms within countries. In Section 4 we will discuss models with more limited international diffusion of knowledge. In these models imperfect diffusion means differences in TFP can play a role in explaining differences in income levels and growth rates. We stress that the Mankiw et al. model relies on even stronger externalities than the typical model of international technology spillovers, such as Parente and Prescott (1994) or Barro and Sala-i-Martin (1995, Chapter 8). We will discuss these models at greater length in Section 4, when we calibrate a hybrid version of them. 2.2. Models with knowledge externalities and new-good externalities Models with both knowledge externalities and new-good externalities are the most plentiful in the endogenous growth literature. By “new-good externalities” we mean surplus to consumers and/or firms from the introduction of new goods. The new goods take the form of new varieties and/or higher quality versions of existing varieties. In Stokey (1988), learning by doing leads to the introduction of new goods over time. The new goods are of higher quality, and eventually displace older goods. The learning is completely external to firms, and what is learned applies to new goods even more than older goods. Hence learning externalities are at the heart of her growth process. In Stokey
822
P.J. Klenow and A. Rodríguez-Clare
(1991), intergenerational human capital externalities (the young learn from the old) are critical for human capital accumulation. Human capital accumulation, in turn, facilitates the introduction of higher quality goods, which are intensive in human capital in her model. Quality ladder models – pioneered by Grossman and Helpman (1991, Chapter 4) and Aghion and Howitt (1992, 1998) – feature knowledge spillovers in that each quality innovation is built on the previous leading-edge technology. Such intertemporal knowledge spillovers are also fundamental in models with expanding product variety, such as Romer (1990) and Grossman and Helpman (1991, Chapter 3). In Romer (1990), A x(i)1−α−β di, Y = HYα Lβ (2.6) A˙ = BHA A.
0
(2.7)
Intermediate goods, the x(i)’s, are imperfect substitutes in production. This is the Dixit– Stiglitz “love of variety” model. The stock of varieties, or ideas, is A. In (2.7) new ideas are invented using human capital and, critically, the previous stock of ideas. This is the intertemporal knowledge spillover. Jones (1995, 2002) argues that, in contrast to (2.7), there are likely to be diminishing returns to the stock of ideas (an exponent less than 1 on A). He bases this on the fact that the number of research scientists and engineers have grown in the U.S. and other rich countries since 1950, yet the growth rate has not risen, as (2.7) would predict. Intertemporal knowledge spillovers still play a pivotal role in Jones’ specification; they are just not as strong as in Romer’s (2.7). More recent models, such as Eaton and Kortum (1996) and Howitt (1999, 2000), continue to emphasize both knowledge externalities and new-good externalities. We will elaborate on these in Section 4 below. 2.3. Models with new-good externalities It is hard to find a model with new-good externalities but without knowledge externalities. We have identified three papers in the literature featuring such models, but two of the papers also have versions of their models with knowledge externalities. Rivera-Batiz and Romer (1991) present a variation on Romer’s (1990) model, as part of their analysis of the potential growth gains from international integration. In their twist, new intermediate goods are invented using factors in the same proportions as for final goods production in (2.6): A x(i)1−α−β di. A˙ = BH α Lβ (2.8) 0
They call this the “lab equipment model” to underscore the use of equipment in the research lab, just like in the production of final goods. In this formulation, they emphasize, “Access to the designs for all previous goods, and familiarity with the ideas and know-how that they represent, does not aid the creation of new designs” (pp. 536– 537). I.e., there are no knowledge externalities, domestic or international. Production of
Ch. 11:
Externalities and Growth
823
ideas is not even knowledge-intensive. Ideas are embodied in goods, however, and there is surplus to downstream consumers from their availability. Rivera-Batiz and Romer note that this model allows countries to benefit from ideas developed elsewhere simply by importing the resulting products. Just as important, international trade allows international specialization in research. Countries can specialize in inventing different products, rather than every product being invented everywhere. In a similar spirit, Romer (1994) considered a model in which knowledge about how to produce different varieties does not flow across countries, but each country can import the varieties that other countries know how to produce. For a small open economy, Romer posited Yt = A
Mt
xjαt
Nt1−α ,
0 < α < 1.
(2.9)
j =1
xj represents the quantity of imports of the j th variety of intermediate good. Because α < 1, intermediate varieties are imperfectly substitutable in production. Firms in the importing country will have higher labor productivity the more import varieties they can access. If exporters cannot perfectly price discriminate and there is perfect competition among domestic final-goods producers, the higher labor productivity (higher Y/N) will benefit domestic workers/consumers. If consumer varieties were imported as well, there would be an additional source of consumer surplus from import varieties. Romer analyzed the impact of import tariffs on the number of varieties M imported in the presence of fixed costs of importing each variety in each country. Although Romer’s model is static, growth in the number of varieties over time, say due to domestic population growth or falling barriers to trade, would be a source of growth in productivity and welfare in his model. Kortum (1997) develops a model in which researchers draw techniques of varying efficiency levels from a Poisson distribution. Kortum does consider spillovers in the form of targeted search. But he also considers the case of blind search, wherein draws are independent of the previous draws. (Kortum fixes the set of goods produced but allows endogenous research into discovering better techniques for producing each good.) In the case of blind search, there are no knowledge spillovers. Growth is sustained solely because of population growth that raises the supply of and demand for researchers. It takes more and more draws to obtain a quality deep enough into the right tail to constitute an improvement. A constant population growth rate sustains a constant flow of quality improvements and, hence, a constant growth rate of income. 2.4. Models with no externalities The seminal growth models without externalities are the AK models of Jones and Manuelli (1990) and Rebelo (1991). In the next section we will present evidence at odds with such models, so we dwell on their implications here. We consider a version close
824
P.J. Klenow and A. Rodríguez-Clare
to Rebelo’s. Final output is a Cobb–Douglas function of physical and human capital: Ct + It = Yt = AK αY t HY1−α t ,
(2.10)
where KY and HY represent the stocks of physical capital and human capital devoted to producing current output. As shown, current output can be used for either consumption or investment. The accumulation equations for physical and human capital are, respectively, KY t+1 + KH t+1 = Kt+1 = (1 − δK )Kt + It , HY t+1 + HH t+1 = Ht+1 =
γ 1−γ (1 − δH )Ht + BH H t KH t .
(2.11) (2.12)
HH and KH represent the stocks of human and physical capital, respectively, devoted to accumulating human capital. We will focus on an equilibrium with a constant fraction of output invested in physical capital (sI = I /Y ) and a constant share of human capital deployed in human capital accumulation (sH = HH /H ). We assume that the ratio of marginal products of physical and human capital are equated across the final output and human capital sectors, so that physical capital is devoted to sK = KH /K =
sH (1 − γ )(1 − α) . sH (1 − γ )(1 − α) + (1 − sH )γ α
(2.13)
The balanced growth rate is defined as 1 + g = Yt+1 /Yt = Kt+1 /Kt = Ht+1 /Ht .
(2.14)
The level of the balanced growth rate is an implicit function of the investment rates and parameter values: (g + δK )1−γ (g + δH )1−α 1−γ γ 1−γ 1−α BsH sK = A(1 − sK )α (1 − sH )1−α sI .
(2.15)
Provided α < 1, human capital is the engine of growth. The growth rate is monotonically increasing in sI because physical capital is an input into human capital accumulation whereas consumption is not. Related, the growth rate does not monotonically increase with the share of inputs devoted to producing human capital. This is because devoting more resources to producing current output increases the stock of physical capital, which is an input into human capital accumulation and hence growth.1 When we look at the data in Section 3, however, we will find no country so high an sH or sK as to inhibit its growth according to this model.
1 To reinforce intuition, consider the (unrealistic) case of γ = 0, wherein new human capital is produced only with physical capital. In this case, growth is not strictly increasing in sK (the share of capital devoted to human capital production) because some physical capital itself needs to be devoted to its own production.
Ch. 11:
Externalities and Growth
825
When α = 1 we have a literal Y = AK model, and the growth rate is solely a function of the physical capital investment rate: g + δK = AsI .
(2.16)
Here there is no point in devoting effort to producing human capital, so sH = 0 and sK = 0. In the special case γ = 1, human capital is produced solely with human capital. This might be called a BH model. Presuming α < 1 of course, the growth rate is simply g + δH = BsH .
(2.17)
Unlike when γ < 1, the growth rate here is monotonically increasing in the effort devoted to adding more human capital. Lucas (1988) and many successors focus on this BH model because human capital accumulation is evidently intensive in human capital. Moreover, even AK models such as Jones and Manuelli (1990) construe their K to incorporate both human capital and physical capital. The consensus for diminishing returns to physical capital (α < 1) is strong. Constant returns are entertained only for a broad measure of physical and human capital. We stress (2.15), a hybrid of AK and BH models, because this generalization allows us to take into account the combined impact of physical and human capital investment rates on growth when physical capital is an input to human capital accumulation (γ < 1). 3. Cross-country evidence In this section we document a number of facts about country growth experiences over the last fifty years. We show that country growth rates appear to depend critically on the growth and income levels of other countries, rather than solely on domestic investment rates in physical and human capital. Cross-country externalities are a promising explanation for this interdependence. In brief, here are the main facts we will present: • The growth slowdown that began in the mid-1970s was a world-wide phenomenon. It hit both rich countries and poor countries, and economies on every continent. • Richer OECD countries grew much more slowly from 1950 to around 1980, despite the fact that richer OECD economies invested at higher rates in physical and human capital. • Differences in country investment rates are far more persistent than differences in country growth rates. • Countries with high investment rates tend to have high levels of income more than they tend to have high growth rates. 3.1. The world-wide growth slowdown As has been widely documented for rich countries, the growth rate of productivity slowed beginning in the early 1970s.2 Less widely known is that the slowdown has 2 The causes of the slowdown remain largely a mystery. For example, see Fischer (1988).
826
P.J. Klenow and A. Rodríguez-Clare Table 2 Output growth declined sharply worldwide Average Y /L growth 1960–75 1975–00
Average sI
Average sH
# of 1960–75 1975–00 # of 1960–75 1975–00 # of countries countries countries
World OECD
2.7% 3.4
1.1% 1.8
96 23
15.8% 23.2
15.5% 22.9
96 23
7.1% 11.4
9.7% 14.3
74 21
Non-OECD
2.5
0.9
73
13.5
13.2
73
5.4
8.0
53
Africa Asia Europe North America South America
2.0 3.2 3.8 2.8 2.3
0.5 2.8 1.9 0.4 −0.1
38 17 18 13 10
12.3 14.5 24.9 14.3 17.3
10.5 19.9 23.1 14.5 15.0
38 17 18 13 10
3.9 6.9 10.7 7.5 7.1
6.0 9.9 13.7 10.2 9.8
19 16 16 13 10
1st quartile (poorest) 2nd quartile 3rd quartile 4th quartile (richest)
1.6 2.6 3.5 3.0
0.5 1.4 1.1 1.5
24 24 24 24
9.6 14.8 15.4 23.6
9.9 14.2 16.3 21.9
24 24 24 24
3.1 5.7 7.5 12.3
5.0 8.9 10.3 15.1
19 19 18 18
Notes: Y /L is GDP per worker. sI is the physical capital investment rate, and sH years of schooling attainment (for the 25+ population) divided by 60 years (working life). Data sources: Barro and Lee (2000) and Heston, Summers and Aten (2002).
been a world-wide phenomenon, rather than just an OECD-specific event.3 We document this in Table 2. Across 96 countries, the growth rate in PPP GDP per worker fell from 2.7% per year over 1960–1975 to 1.1% per year over 1975–2000. Growth decelerated 1.6 percentage points on average in both the sample of 23 OECD countries and the in the sample of 73 non-OECD countries.4 The slowdown hit North and South America the hardest (their growth rates fell 2.4 percentage points) and barely brushed Asia (who slowed down just 0.4 of a percentage point). The slowdown hit all income quartiles of the 96 country sample (based on PPP income per worker in 1975). Although each income quartile grew at least one percentage point slower, the slowdown was not as severe in the poorest half as in the richest half. China’s growth rate actually accelerated from 1.8 to 5.1, in the wake of reforms that began in the late 1970s. Chile, which experienced rapid growth in the 1990s, accelerated 2.1 percentage points. Why does a world-wide growth slowdown suggest international externalities? Couldn’t it simply reflect declining investment rates world-wide, as suggested by the AK model in the previous section? Table 2 also shows what average investment rates in 3 An exception is Easterly (2001b). 4 OECD countries are based on 1975 membership. There were 24 OECD members in 1975, but the Penn
World Tables contain data for unified Germany only back to 1970.
Ch. 11:
Externalities and Growth
827
physical and human capital did before and after the mid-1970s. The investment rates in physical capital come from Penn World Table 6.1. As a proxy for the fraction of time devoted to accumulating more human capital, we used years of schooling attainment relative to a 60-year working life. We used data on schooling attainment for the 25 and older population from Barro and Lee (2000). This human capital investment rate, which averages around 7% across countries, reflects the fraction of ages 5 to 65 devoted to schooling as opposed to working. We prefer the attainment of the workforce as opposed to the enrollment rates of the school-age population. The latter should take a long time to affect the workforce and therefore the growth rate. According to Table 2, the average investment rate in physical capital across all countries was virtually unchanged (15.8% before vs. 15.5% after the slowdown), and the investment rate in human capital actually rose strongly (going from 7.1% to 9.7%). The same pattern applies for the OECD and non-OECD separately, and for all four quartiles of initial income. Thus the growth slowdown cannot be attributed to a world-wide decline in investment rates. The breadth of the growth slowdown suggests something linking country growth rates, and ostensibly something other than investment rates.5 This is contrary to the predictions of AK models, in which the growth rate of a country depends on domestic investment rates. The world-wide nature of the slowdown suggests that endogenous growth models, more generally, should not be applied to individual countries but rather to a collection of interdependent countries. Knowledge diffusion through trade, migration, and foreign direct investment are likely sources of interdependence. Three other examples of interdependence are offered by Parente and Prescott (2005). First, growth rates picked up in the 20th century relative to the 19th century for many countries. Second, the time it takes a country to go from $2000 to $4000 in per capita income has fallen over the 20th century, suggesting the potential to grow rapidly by adopting technology in use elsewhere. Third and related, they stress that “growth miracles” always occur in countries with incomes well beneath the richest countries, again consistent with adoption of technology from abroad. Knowledge diffusion, broadly construed, could include imitation of successful institutions and policies in other countries. Kremer, Onatski and Stock (2001) argue that such imitation might explain the empirical transition matrix of the world income distribution. If improving institutions leads only to static gains in efficiency, however, then the barriers to imitation have to be large to explain why the best institutions are not in place everywhere. As we will illustrate in Section 4 below, the required barriers to technology adoption are modest precisely because the benefits accumulate with investments. 3.2. Beta convergence in the OECD As documented by Baumol (1986) and many others, incomes have generally been converging in the OECD. Barro and Sala-i-Martin (1995) used the term sigma con5 It also casts doubt on explanations for the growth slowdown that are confined to rich countries.
828
P.J. Klenow and A. Rodríguez-Clare
Figure 1. OECD incomes correlate negatively with growth rates, positively with investment rates. Source: Penn World Table 6.1 data on 23 OECD Countries.
vergence to describe such episodes of declining cross-sectional standard deviations in log incomes. We focus on a related concept that Barro and Sala-i-Martin labeled beta convergence, namely a negative correlation between a country’s initial income level and its subsequent growth rate. We look at beta convergence year by year in Figure 1. The data on PPP income per worker comes from Penn World Table 6.1 [Heston, Summers and Aten (2002)], and covers 23 OECD countries over 1960–2000. The figure shows the correlation between current income and growth hovering between −0.50 and −0.75 from 1960 through the early 1980s. The correlation was still negative from the mid1980s through the mid-1990s, but less so, and turned positive in the latter 1990s. DeLong (1988) pointed out that a country’s OECD membership is endogenous to its level of income, so that members at time t will tend to converge toward each other’s incomes leading up to time t. Our focus, however, is not on convergence per se. Our point is instead about how investment rates correlate with income during the period of convergence. Figure 1 also shows the physical capital investment rate, and it is positively correlated with a country’s income throughout the sample. Figure 2 shows that schooling attainment is also positively correlated with income throughout the sample. How do these investment correlations square with AK-type models with no externalities? Expression (2.15) shows that a country’s growth rate should be increasing in its investment rates. For beta convergence to occur in this model, a country’s investment rates must be negatively correlated with a country’s level of income. But Figures 1 and 2 show the opposite is true: in every year, richer OECD countries had higher investment rates in human and physical capital than poorer OECD countries did. According to this class of models, OECD countries should have been diverging throughout the entire sample, rather than converging through most of it. Now, this reasoning ignores likely
Ch. 11:
Externalities and Growth
829
Figure 2. OECD incomes correlate negatively with growth rates, positively with investment schooling. Source: Penn World Table 6.1 and Barro and Lee (2000) data for 21 OECD Countries.
differences in efficiency parameters A and B across countries. But rescuing AK models would require that richer countries have lower efficiency parameters. We would guess that rich countries tend to have better rather than worse institutions [e.g., Hall and Jones (1999)]. 3.3. Low persistence of growth rate differences Easterly et al. (1993) documented that country growth rate differences do not persist much from decade to decade. They estimated correlations of around 0.1 to 0.3 across decades. In contrast, they found that country characteristics such as education levels and investment rates exhibit cross-decade correlations in the 0.6 to 0.9 range. Just as we do, they suggest country characteristics may determine relative income levels and world-wide technological changes long-run growth. Easterly and Levine (2001) similarly provide evidence that “growth is not persistent, but factor accumulation is”. In Table 3 we present similar findings. We compare average growth rates from 1980– 2000 vs. 1960–1980, and from decade to decade within 1960–2000. We find growth rates much less persistent than investment rates for the world as a whole, and for the OECD and non-OECD separately. Again, these facts seem hard to reconcile with the AK model in which a country’s domestic investment rates determine its growth rate. Figure 3 illustrates a related pattern: deciles of countries (based on 1960 income per worker) grew at similar average rates from 1960 to 2000. Each decile consists of the unweighted average of income per worker in 9 or 10 countries. The average growth rate is 1.7% in the sample, and the bottom decile in 1960 grew at precisely this rate. This figure suggests movements in relative incomes, but no permanent differences in
830
P.J. Klenow and A. Rodríguez-Clare Table 3 Investment rates are more persistent than growth rates 1980–2000 vs. 1960–1980
Decade to decade
Y /L growth
sI
sH
Y /L growth
sI
sH
World
0.34 (0.13)
0.56 (0.07)
1.02 (0.04)
0.20 (0.07)
0.77 (0.04)
1.00 (0.02)
OECD
0.12 (0.13)
0.44 (0.09)
0.86 (0.08)
0.27 (0.09)
0.70 (0.06)
0.92 (0.03)
Non-OECD
0.36 (0.17)
0.44 (0.09)
1.10 (0.07)
0.17 (0.08)
0.71 (0.05)
1.04 (0.03)
Notes: World = 74 countries with available data; OECD = 22 countries; and non-OECD = 52 countries. Decades consisted of the 1960s, 1970s, 1980s, and 1990s. All variables are averages over the indicated periods. Each entry is from a single regression. Bold entries indicate p-values of 1% or less. Data Sources: Barro and Lee (2000) and Penn World Table 6.1 [Heston, Summers and Aten (2002)].
Figure 3. The evolution of income for 1960 deciles. Source: Penn World Table 6.1.
long-run growth rates, even comparing the richest and poorest countries. This sample contains 96 countries, and therefore many of the poorest countries mired in zero or negative growth. Pritchett (1997), on the other hand, offers compelling evidence that incomes diverged massively from 1800 to 1960. Doesn’t this divergence favor models, such as AK with-
Ch. 11:
Externalities and Growth
831
Table 4 Investment rates correlate more with levels than with growth rates Independent variable = sI
Independent variable = sH
Dependent variable
Dependent variable
Y /L Growth rates
Y /L Log levels
# of countries
Y /L Growth rates
Y /L Log levels
# of countries
All countries
0.111 (0.017) R2 = 0.32
1.25 (0.13) R2 = 0.48
96
0.210 (0.060) R2 = 0.15
0.313 (0.026) R2 = 0.67
74
OECD
0.020 (0.047) R2 = 0.01
0.760 (0.358) R2 = 0.18
23
−0.259 x (0.078) R2 = 0.37
0.119 (0.024) R2 = 0.56
21
Non-OECD
0.124 (0.023) R2 = 0.29
0.842 (0.162) R2 = 0.28
73
0.367 (0.095) R2 = 0.22
0.314 (0.043) R2 = 0.51
53
Notes: Variables are averages over 1960–2000. Each entry is from a single regression. Bold entries indicate p-values of 1% or less. Data sources: Barro and Lee (2000) and Penn World Table 6.1 [Heston, Summers and Aten (2002)].
out international externalities, in which country growth rates are not intertwined? Not necessarily. As argued by Parente and Prescott (2005), the opening up of large income differences coincided with the onset of modern economic growth. The divergence could reflect the interaction of country-specific barriers to technology adoption with the emergence of modern technology-driven growth. More generally, any given divergence episode could reflect widening barriers to importing technology rather than simply differences in conventional investment rates. 3.4. Investment rates and growth vs. levels The AK model we sketched in the previous section predicts that a country’s growth rate will be strongly related to its investment rates in physical and human capital. In Table 4 we investigate this empirically in cross-sections of countries over 1960–2000. In four of the six cases, the average investment rate is positively and significantly related to the average growth rate. For the OECD, the physical capital investment rate is not significantly related to country growth, and the human capital investment rate is actually negatively and significantly related to country growth. But for the non-OECD and allcountry samples, the signs and significance are as predicted. This evidence constitutes the empirical bulwark for AK models. In the four cases where the signs are as predicted, are the magnitudes roughly as an AK model would predict? First consider the literal AK model. According to (2.16) in the previous section, the coefficient on sI should be A. What might be a reasonable
832
P.J. Klenow and A. Rodríguez-Clare
value for A? In order to match the average growth rate in GDP per worker (1.8%), given an average investment rate in physical capital (17%) and a customary depreciation rate (8%), the value of A would need to be A=
0.018 + 0.08 ∼ g avg + δK = = 0.58. avg 0.17 sI
(3.1)
This level of A is more than four times larger than the two significant positive coefficients on sI in the first column of Table 4, which are around 0.12. The estimated coefficients are small in magnitude compared to what an AK model would predict. This discrepancy could reflect classical measurement error in investment rates, but such measurement error would need to account for more than 80% of the variance of investment rates across countries. Plus one would expect positive endogeneity bias in estimating the average level of A, due to variation in A across countries that is positively correlated with variation in sI . We next consider the literal BH model. According to (2.17), the coefficient on sH should be B. To produce the average growth rate in GDP per worker given the average investment rate in human capital (8.8%) and a modest depreciation rate (2%), B would need to be B=
g avg + δH 0.018 + 0.02 ∼ = = 0.43. avg .088 sH
(3.2)
The third column of estimates in Table 4 contain coefficients on sH . Of the two positive coefficients, one is half the predicted level (0.21) whereas the other is not far from the predicted level (0.37). Finally, consider the hybrid model in (2.15). We assume γ = 0.9 so that human capital accumulation is intensive in human capital, but does use some physical capital. For producing current output we assume the standard physical capital share of α = 1/3. We set the depreciation rates as previously mentioned. We set sK , the share of physical capital devoted to human capital accumulation, based on (2.13). As (2.15) illustrates, we cannot independently identify A and B, only their product. We set A1−γ B 1−α ∼ = 0.60, so that the average predicted growth rate from (2.15) and observed sH and sI investment rates matches the average growth rate in GDP per worker of 1.8%. We then regress actual growth rates on predicted growth rates for a cross-section of 73 countries with available data. The coefficient estimated is 0.26 (standard error 0.08, R2 of 0.13), far below the theoretical value of 1. Again, the empirical estimate might be low because of measurement error in predicted growth, but it would need to be large. To recap, only 1 of the 7 coefficients of growth on investment rates considered is in the ballpark of an AK model’s prediction. In contrast, we obtain uniformly positive and significant coefficients when we regress (log) levels of country income on country investment rates. In 5 of the 6 cases, the R2 is notably higher with levels than with growth rates. Investment rates appear far better at explaining relative income levels than relative growth rates. The driver of growth rates would appear to be something other than simply domestic investment rates.
Ch. 11:
Externalities and Growth
833
The preceding discussion focused on the steady-state predictions of AK models. It is possible that AK models fare better empirically when transition dynamics are taken into account. But it is worth noting that Klenow and Rodríguez-Clare (1997), Hall and Jones (1999), Bils and Klenow (2000), Easterly (2001a), Easterly and Levine (2001), and Hendricks (2002) all find that no more than half of the variation in growth rates or income levels can be attributed directly to human and physical capital. Pritchett (2005), who considers many different parameterizations of the human capital accumulation technology, likewise finds that human capital does not account for much cross-country variation in growth rates. 3.5. R&D and TFP We now turn away from AK models to a model with diminishing returns to physical and human capital, but with R&D as another form of investment. Such a model might be able to explain country growth rates with no reference to cross-country externalities. For example, perhaps a variant of the Romer (1990) model could be applied country by country, with no international knowledge flows. R&D investment would have to behave in a way that leads to a worldwide growth slowdown, beta convergence in the OECD, and low persistence of growth rate differences. And, more directly, R&D investment would have to explain country growth rates. Research effort, like human capital, is difficult to measure. But Lederman and Saenz (2003) have compiled data on R&D spending for many countries. We now ask the same questions of their R&D investment rates that we asked of investment rates in physical and human capital: how correlated are R&D investment rates with country growth rates and country income levels? The first column in Table 5 says that countries with high R&D spending relative to GDP do not grow systematically faster.6 Countries with high R&D shares do, however, tend to have high relative incomes. But the correlation with income is not significant outside the OECD. One possibility is that these regressions do not adequately control for the contributions of physical and capital. We therefore move to construct Total Factor Productivity (TFP) growth rates and levels. We subtract estimates of human and physical capital per worker from GDP per worker: ln TFP = ln(Y/L) − α ln(K/L) − (1 − α) ln(H /L)
(3.3)
where Y is real GDP, L is employment, K is the real stock of physical capital, and H is the real stock of human capital. We suppress time and country subscripts in (3.3) for readability. We would prefer to let α vary across countries and across time based on factor shares, but such data is not readily available for most countries in the sample. We instead set α = 1/3 for all countries and time periods. Gollin (2002) finds that capital’s 6 Because R&D data was not available for all country-years between 1960 and 2000, we took time effects out of the variables (growth rates, income levels, investment rates in R&D), then averaged the residuals over time.
834
P.J. Klenow and A. Rodríguez-Clare Table 5 R&D intensity also correlates more with levels than growth rates Independent variable = R&D spending as a share of GDP Dependent variable
Dependent variable
Y /L Growth rates
Y /L Log levels
# of countries
TFP Growth rates
TFP Log levels
# of countries
All countries
0.40 (0.59) R2 = 0.01
0.69 (0.23) R2 = 0.10
82
0.43 (0.52) R2 = 0.01
0.37 (0.08) R2 = 0.27
67
OECD
−0.15 (0.46) R2 = 0.01
0.42 (0.11) R2 = 0.45
21
−0.16 (0.32) R2 = 0.01
0.17 (0.06) R2 = 0.28
21
Non-OECD
0.88 (1.03) R2 = 0.01
0.55 (0.41) R2 = 0.03
61
0.85 (1.01) R2 = 0.02
0.34 (0.14) R2 = 0.12
46
Notes: Variables are country averages over years in 1960–2000 with data relative to time effects. Y /L is GDP per worker. TFP nets out contributions from human and physical capital, as described in the text. Each entry is from a single regression. Bold entries indicate p-values of 2% or less. Data sources: Barro and Lee (2000), Penn World Table 6.1 [Heston, Summers and Aten (2002)], and Lederman and Saenz (2003).
share varies from 0.20 to 0.35 across a sample of countries, but does not correlate with country income levels or growth rates. We use Penn World Table 6.1 data assembled by Heston, Summers and Aten (2002) for PPP GDP, employment, and PPP investment in physical capital. We assume an 8% geometric depreciation rate and the usual accumulation equation to cumulate investment into physical capital stocks. We approximate initial capital stocks using the procedure in Klenow and Rodríguez-Clare (1997, p. 78). We let human capital per worker be a simple Mincerian function of schooling: H = hL = exp(φs)L.
(3.4)
Here h represents human capital per worker, and s denotes years of schooling attainment. We use Barro and Lee (2000) data on the schooling attainment of the 25 and older population. This data is available every five years from 1960 to 2000, with the last year an extrapolation based on enrollment rates and the slow-moving stock of workers. A more complete Mincerian formulation would include years of experience in addition to schooling and would sum the human capital stocks of workers with different education and experience levels. In Klenow and Rodríguez-Clare (1997) we found that taking experience and heterogeneity into account had little effect on aggregate levels and growth rates, so we do not pursue it here. We use (3.4) with the Mincerian return φ = 0.085, based on the returns estimated for many countries and described by Psacharopoulos and Patrinos (2002).
Ch. 11:
Externalities and Growth
835
The latter columns in Table 5 present regressions of TFP growth rates and levels on R&D investment rates. The sample of countries is smaller given data limitations (67 countries rather than 82). Just like growth in GDP per worker, growth in TFP is not significantly related to R&D investment rates. But TFP levels, like levels of GDP per worker, are positively and significantly related to R&D investment rates. From this we conclude that even R&D investment rates affect relative income levels, not longrun growth rates. The persistence of R&D investment rate differences across countries, combined with the lack of persistent growth rate differences, supports this interpretation. We are led to consider models in which country growth rates are tethered together. Before considering a model with international knowledge externalities, we pause to consider a model with “externalities” operating through the terms of trade. We have in mind Acemoglu and Ventura’s (2002) model of the world income distribution. In their model, each country operates an AK technology, but uses it to produce distinct national varieties. Countries with high AK levels due to high investment rates plentifully supply their varieties, driving down their prices on the world market. This results in a pAK model with a stationary distribution of income even in the face of permanent differences in country investment rates (and A levels, for that matter). Prices tether incomes together in the world distribution, not the flow of ideas. This is a clever and coherent model, but we question its empirical relevance. Hummels and Klenow (2005) find that richer countries tend to export a given product at higher rather than lower prices. They do estimate modestly lower quality-adjusted prices for richer countries, but nowhere near the extent needed to offset AK forces and generate “only” a factor of 30 difference in incomes. To summarize this section, AK models tightly connect investment rates and growth rates. Such a tight connection does not hold empirically. This is the case for the world growth slowdown, for OECD convergence, for growth persistence, and for country variation in growth vs. income levels. A version of the AK model with endogenous terms of trade might be able to circumvent these empirical hazards but faces empirical troubles of its own. We therefore turn to models with international knowledge externalities that drive long-run growth.
4. Models with common growth driven by international knowledge spillovers Based on evidence in the previous section, we now focus on models with two features. The first is that, in steady state, all countries grow at the same rate thanks to international knowledge spillovers. The second feature is that differences in policies or other country parameters generate differences in TFP levels rather than growth rates. Examples of this type of model are Howitt (2000), Parente and Prescott (1994), and Eaton and Kortum (1996, 1999), as well as the model of technology diffusion in Chapter 8 of Barro and Sala-i-Martin (1995). Eaton and Kortum (1999) is a particularly close predecessor to our approach, although they go farther in tying their analysis to international patenting data in the G5 countries.
836
P.J. Klenow and A. Rodríguez-Clare
In these models there is a world technology frontier, and a country’s research efforts determine how close the country gets to that frontier. There are three different issues that must be addressed. First, what determines the growth rate of the world technology frontier? Second, how is it that a country’s research efforts allow it to “tap into” the world technology frontier? And third, what explains differences across countries in their research efforts? Our goal in this section is to build on the ideas developed in the recent literature to construct a model that offers a unified treatment of these three issues and that is amenable to calibration. The calibration is intended to gauge the model’s implications about the strength of the different externalities and the drivers of cross-country productivity differences.7 To highlight the different issues relevant for the model, our strategy is to present it in parts. The next subsection 4.1 takes world growth and R&D investment as exogenous and discusses how R&D investment determines steady state relative productivity. Subsection 4.2 discusses different ways of modeling how world-wide R&D investment determines the growth rate of the world technology frontier. Subsection 4.3 extends the model so as to allow for endogenous determination of countries’ R&D investment rates. Subsection 4.4 calibrates the model. Finally, subsection 4.5 presents the results of an exercise where we calculate, for each country in our sample, the impact on productivity from international spillovers. 4.1. R&D investment and relative productivity In this section we focus on a single country whose research efforts determine its productivity relative to the world technology frontier. Both the R&D investment rate and the rate of growth of the world technology frontier are exogenous. Output is produced with a Cobb–Douglas production function: Y = K α (AhL)1−α , where Y is total output, K is the physical capital stock, A is a technology index, h is human capital per person, and L is the total labor force. We assume that h is constant and exogenous. Output can be used for consumption (C), investment (I ), or research (R), Y = C + pI + R, where p is the relative price of investment and is assumed constant through time. Capital is accumulated according to: K˙ = I − δK. Finally, A evolves according to: A˙ = (λR/L + εA) 1 − A/A∗ (4.1) where λ is a positive parameter and A∗ is the world technology frontier, both common across countries.8 There are three salient differences between this model and the standard endogenous growth model. Firstly, the productivity of research in generating A-growth is affected by 7 Although we refer to research externalities throughout this section, the knowledge externalities could just as well be with respect to human capital. Only when we use data on R&D is the analysis specific to research. 8 In models like those of Parente and Prescott (1994) and Howitt (2000) research is meant to capture both R&D and technology adoption efforts. In this paper we follow this practice and simply refer to the sum of these two technology investments as R&D or just “research”.
Ch. 11:
Externalities and Growth
837
the country’s productivity relative to the frontier, as determined by the term (1 − A/A∗ ) in (4.1). This captures the idea that there are “benefits to backwardness”. One reason for this may be that the effective cost of innovation and technology adoption falls when a country is further away from the world technology frontier. This is what happens in Parente and Prescott (1994) and in Barro and Sala-i-Martin (1995, Chapter 8). Alternatively, being further behind the frontier may confer an advantage because every successful technology adoption entails a greater improvement in the national technology level. This is what happens in Howitt (2000) and in Eaton and Kortum (1996).9 Secondly, we introduce ε 0 to capture the sources of technology diffusion from abroad that do not depend on domestic research efforts. We have in mind imports of goods that embody technology, and that do not require upfront adoption costs (e.g., equipment which is no harder to use but which operates more efficiently).10 As we will see below, this is important for the model to match certain features of the data. Thirdly, in contrast to most endogenous growth models, we divide research effort by L in the A-growth expression above. This is done to get rid of scale effects and can be motivated in two ways. First, if A represents the quality of inputs, then one can envisage a process where an increase in the labor force leads to an expansion in the variety of inputs [Young (1998) and Howitt (1999)]. With a larger variety of inputs, research effort per variety is diluted. This eliminates the impact of L on A growth. Second, if research is undertaken by firms to increase their own productivity, then population growth may lead to an expansion in the number of firms and a decrease in the impact of aggregate research on firms’ A-growth [Parente and Prescott (1994)]. In this case, L represents the number of firms. The measured R&D investment rate is given by sR = R/Y . This implies that R/(AL) = sR Y/(AL) = sR k where k ≡ (K/Y )α/(1−α) h = Y/(AL). To proceed, note that in steady state a ≡ A/A∗ will be constant, since A will grow at the same rate as A∗ , which we denote by gA . Thus, from (4.1) gA = (λsR k + ε)(1 − a). Solving for a we obtain: gA . a =1− λsR k + ε
(4.2)
(4.3)
The values of k and sR determine a country’s relative A from (4.3). Conceivably, the parameter λ (TFP in research, if you will) could differ across countries and also contribute to differences in A. But in this paper we assume λ does not vary across countries. We 9 In Howitt’s model, (1 − A/A∗ ) arises from the product of two terms: (1/A∗ )(A∗ − A). The (1/A∗ ) term
arises because, as the world’s technology becomes more advanced, more research is required to tap into it; the second term captures the fact that, when the country is more backward, every successful technology adoption entails a greater improvement in the national technology level. 10 This free flow of ideas is also likely to depend on the local presence of multinationals, which bring valuable knowledge that diffuses to other local firms without the need for additional R&D.
838
P.J. Klenow and A. Rodríguez-Clare
do, however, allow researchers to be more productive in countries with more physical and human capital per worker. The previous results clearly show that policies that lower investment in physical or human capital or R&D do not affect a country’s growth rate. Their effect is on a country’s steady state relative A. Also, as discussed above, there are no scale effects in this model: higher L does not lead to higher growth or to a higher relative A. This stands in contrast to most growth models based on research [e.g., Romer (1990), Barro and Sala-i-Martin (1995, Chapter 8)]. It is also noteworthy in Equation (4.3) that the value of k, which captures physical and human capital intensity, affects a country’s TFP level conditional on its R&D investment rate. Thus, large differences in TFP across countries do not necessarily imply that differences in human and physical capital stocks are just a small part of cross country income differences. Indeed, this model suggests that some of the TFP differences may be due to differences in capital intensities across countries. Below we explore this issue quantitatively. It is instructive to calculate the social rate of return to research at the national level. As shown in Jones and Williams (1998), this can be done even without knowing the details of the model that affect the endogenous determination of the R&D investment rate. Letting A˙ = G(A, R), Jones and Williams show that the (within-country) social rate of return r˜ can be expressed as: r˜ =
∂Y/∂A + ∂G/∂A + gPA . PA
(4.4)
Here PA stands for the price of ideas and is given by PA = (∂G/∂R)−1 . As explained by Jones and Williams, the first two terms in (4.4) represent the dividends of research while the third term represents the associated capital gains. The first dividend term is the obvious component, namely the productivity gain from an additional idea divided by the price of ideas. The second dividend term captures how an additional idea affects the productivity of future R&D. In the model we presented above, it is straightforward to show that, along a steady state path, we have: agA r˜ = (1 − α)λk(1 − a) + ε(1 − a) − (4.5) + gL . 1−a The first term on the right-hand side corresponds to the first dividend term in Jones and Williams’ formula. The second term, in square brackets, corresponds to the indirect effect of increasing A on the cost of research (∂G/∂A). The third term, gL , corresponds to the term capturing the capital gains in Jones and Williams formula. To understand this last term, note that we have implicitly assumed that new varieties or firms start up with the same productivity as existing varieties or firms. Thus, the value of ideas will rise faster with a higher gL , and the social return to research will correspondingly increase with gL .
Ch. 11:
Externalities and Growth
839
Also note that, since the RHS of (4.5) is decreasing in a, and a is increasing in sR , the social rate of return to research will be decreasing in sR , as one would expect. If k varies less than in the data, one should also expect to find higher social rates of return to research in poor countries than in rich countries, as found by Lederman and Maloney (2003). More importantly, if ε is close to zero, then from (4.2) and (4.5) we should have r˜ ≈ (1 − α)λk(1 − a) ≈ (1 − α)gA /sR . Using the growth rate of A in the OECD in the period 1960–2000 as an approximation of gA (1.5%), and using α = 1/3, then r˜ ≈ 0.01/sR . Noting that the median of sR in the non-OECD countries we have in our sample is sR = 0.5%, then r˜ 200%. This seems implausibly high.11 There are two ways out of this problem. First, one can argue that measured R&D investment does not capture all the research efforts undertaken by countries. Clearly, higher R&D investment rates would lead to lower and more plausible social rates of return to research. Second, one can argue that the implausible implications of the model are due to the assumption that ε is close to zero. In the calibration exercise in Section 4.4, we will argue that both of these solutions are needed to make the model consistent with the data. 4.2. Modeling growth in the world technology frontier In this section we extend the model so that gA is endogenously determined by the research efforts in all countries. The models we mentioned above deal with this in different ways, except Parente and Prescott who leave gA as exogenous. Barro and Sala-i-Martin (1995, Chapter 8) have a Romer-type model of innovation that determines gA in the “North”. We do not pursue this possibility because of the scale effect that arises in their model (larger L in the North leads to higher gA ) and because we want to allow research efforts by all countries to contribute to the world growth rate. We first consider an adaptation of Howitt’s (2000) formulation. A country’s total effective research effort, λRi , gets diluted by the country’s number of varieties or number of firms, both represented by Li , and is then multiplied by a common spillover parameter, σ , to determine that country’s contribution to the growth of the world’s technology frontier: λRi
. A˙ ∗ = σ Li i
Given our results above, we obtain: gA = σ λki sRi ai .
(H1)
i
This formulation has the nice feature that the world growth rate does not depend on the world’s level of L (no scale effect on growth at the world level), although it does depend 11 The problem is not so pronounced for the U.S. Given its measured R&D investment rate of s = 2.5%, we R
have r˜ ≈ 40%, which is in the range of estimates of the social rate of return to R&D in the U.S. See Griliches (1992) and Hall (1996).
840
P.J. Klenow and A. Rodríguez-Clare
positively on R&D investment rates. The main problem with this formulation, and the reason we do not pursue it further, is that larger countries contribute no more to world growth than smaller countries do. This has the implausible implication that subdividing countries would raise the world growth rate. In footnote 22 of his paper, Howitt discusses an alternative specification wherein country spillovers are diluted by world variety rather than each country’s variety. This implies that: λRi
. A˙ ∗ = σ L i where L = Li . Howitt does not pursue this approach because, in the presence of steady-state differences in the rate of growth of L across countries, gA would be completely determined in the limit by the research effort of the country with the largest rate of growth of L. We believe, however, that it is quite natural to analyze the case in which gL is the same across countries.12 In this case, ωi ≡ Li /L is constant through time, and the expression above can be manipulated to yield: gA = σ (H2) λki sRi ai ωi . i
If we think of L as the number of firms rather than the number of varieties of capital goods, then (H2) amounts to stating that gA is determined by the country-workforceweighted average research intensity across firms world-wide. This seems much more reasonable than (H1), where gA is determined by the unweighted average of research intensity across countries. Expression (H2) differs from (H1) only in the presence of the weights ωi that represent shares of world L. This has two advantages: first, large countries contribute more to world growth than small countries do, and second, subdividing countries would not affect the world growth rate. But (H2) has a problematic implication, namely that those countries with higher than average ki sRi ai would be better off disengaging from the rest of the world – their growth rate would be higher if they were isolated. That is, a research intensive country would be better off ignoring the research done in other countries. According to Howitt’s variety interpretation of this model, this is because an isolated country’s growth rate would be given by σ λki sRi ai . Its research intensity would no longer be spread out over the number of world varieties, but instead over the smaller number of the country’s own varieties. Thus, when a country disengages, it no longer benefits from spillovers from research conducted by the rest of the world, but there is 12 If one country’s population did come to dominate world population, however, it might be sensible to say
it does almost all of the world’s research and, hence, it will virtually determine the world growth rate. We assume equal labor force growth rates across countries not because we think it is accurate for describing what is happening now, but because we think it is a convenient fiction for a steady state model to explore international spillovers.
Ch. 11:
Externalities and Growth
841
an important compensating gain that comes from the fact that variety – and therefore dilution – falls for the disengaging country. Since there is no love of variety in Howitt’s model, a high research-intensity country would gain from disengagement. By this logic, engagement could not be sustained among any set of asymmetric countries! The higher ki sRi ai countries would always prefer to disengage, leaving all countries isolated in equilibrium. We now turn to an alternative specification for world spillovers in which variety does not play such a crucial role. The specification will exhibit several of the features we have been looking for: first, no scale effect of world population on the world’s growth rate; second, other things equal, larger countries contribute more to world prosperity than small countries do; and third, tapping into rest-of-world research does not require spreading research across more varieties. We believe this is accomplished by adopting the formulation in Jones (1995): instead of dividing by L, the scale effect is avoided by introducing the assumption that advancing the world technology frontier gets harder as the frontier gets higher. This can be captured by the following specification of international spillovers: γ −1 σ λRi A˙ ∗ = A∗ (4.6) i
where γ < 1. In this setting, sustained growth in A∗ depends on a continuously rising population. To see this, notice that we can restate (4.6) as follows: γ −1 Lσ λki sRi ai ωi . gA = A∗ (J) i
This expression makes clear that gA is decreasing in A∗ ; as mentioned above, this is what is going to eliminate the scale effect. Since all of the terms in the summation on the right-hand side of (J) are constant, then – differentiating with respect to time – we get that: gL gA = (4.7) . 1−γ One criticism of this specification is that gA does not depend on sR , hence policyinduced increases in research intensity would not increase the world’s growth rate [Howitt (1999)]. As Jones (2002) argues, however, research intensity has been increasing over the last decades without a concomitant increase in the growth rate, so it is far from clear that we want a model where gA depends on sR .13 An interesting and relevant feature of the model presented by Eaton and Kortum (1996) is that it allows for spillovers to differ between pairs of countries. We can introduce this feature in the model by doing two things: first, we allow each country to 13 Even though research intensity does not affect the growth rate, it can have sizable effects on welfare,
particularly when – as evidence suggests – the social rate of return of research is significantly higher than the private rate of return.
842
P.J. Klenow and A. Rodríguez-Clare
have a different technology frontier, A∗i ; second, we add country-pair specific spillover parameters, ηil , to (4.6) so that now: γ −1 A˙ ∗i = A∗i σ λRl ηil . l
This formulation implies that there will no longer be a world technology frontier in the way it existed in model (J). However, it proves useful for the analysis to introduce a new concept, which we will denote by A˜ and which could be understood as the “frictionless technology frontier”. To define this concept, note that if spillovers were the same among all country pairs (ηil = 1 for all i and l) – a case we could interpret as frictionless – then countries would have a common technology frontier: A∗i = A∗l for all i and l. We define A˜ so that in this case (ηil = 1 for all i and l) A∗i = A˜ for all i. As we will see ˜ which below, in steady state A˜ grows at the same rate as A∗i for all i. Letting zi ≡ A∗i /A, captures the strength of spillovers from the rest of the world to country i, we arrive at the following steady state restriction: γ −1 gA = A∗i (JEK) L(σ/zi ) λkl sRl al ωl zl ηil l
where JEK stands for Jones, Eaton and Kortum and where al is now country l s technology level relative to its own technology frontier: al ≡ Al /A∗l . It can be shown that ˜ this implies the following restriction for A: A˜ = (vL)1/(1−γ ) (4.8) where v ≡ (σ/gA ) l λkl sRl al ωl . It is clear that each country’s technology frontier and A˜ will grow at the same rate as A∗ did in model (J), given by gL /(1 − γ ). It is interesting to pause here to discuss the model’s implications regarding the effect of country size on productivity. Imagine, to simplify, that all countries are the same except for size, and assume that ηij = 1 for i = j and ηij = η < 1 for i = j . Then it is easy to show that zi > zj if ωi > ωj ; larger countries are more productive. Intuitively, larger countries benefit more from spillovers because more of the world’s research takes place within their borders. As long as borders discretely reduce spillovers, larger countries will capture more spillovers and enjoy higher productivity. The next step is to impose some restrictions on the international spillover parameters. The literature has allowed international spillovers to depend on trade [Coe and Helpman (1995)], distance [Eaton and Kortum (1996)], and other variables such as FDI flows [Caves (1996)]. Here we focus on the simplest approach, which is to assume that the parameters ηil are completely determined by distance. (This would capture trade and FDI related spillovers that are related to distance.) We do this by assuming that ηil = e−θd(i,l) , where d(i, l) is bilateral distance between countries i and l, and θ is some positive parameter. This model collapses to (J) if θ = 0. This completes our discussion of different ways to model international spillovers. Table 6 summarizes the discussion in this subsection.
Ch. 11:
Externalities and Growth
843
Table 6 Alternative ways of modeling international spillovers Spillovers
λRi Li
H1
A˙ ∗ = σ
H2
A˙ ∗ = σ
J
γ −1 A˙ ∗ = A∗ σ i λRi
i
i
λRi L
Growth rate
Advantages
Disadvantages
gA = σ i λki sRi ai
No scale effects
Larger countries contribute no more to gA than do small countries
gA = σ i λki sRi ai ωi where ωi = Li /L
Previous ones plus: Size matters for a country’s contribution to gA
Countries with higher than average ki sRi ai would be better off ignoring research from the rest of the world
gA = gL /(1 − γ ) Previous ones plus: Research-intensive countries do not prefer to disengage from the rest of the world. gA = gL /(1 − γ ) Previous ones plus: The model takes into account effect of distance on spillovers.
gA does not depend on R&D efforts . . . but is this a disadvantage? [See Jones (1995).]
γ −1 σ l λRl ηil JEK A˙ ∗i = A∗i
We will find it hard to see the cost of geographic isolation in the TFP data.
4.3. Determinants of R&D investment We mentioned above that there are two ways to motivate the model we presented in subsection 4.1. First, we can think of a model like the one presented in Howitt (2000), where research leads to improvements in the quality of capital goods, and population growth leads to an expansion in the total number of varieties available. Second, research may be carried out by firms to increase their own productivity, as in Parente and Prescott (1994). We pursue this second approach because it is simpler and much more convenient for our calibration purposes later on. As in Parente and Prescott (1994), we assume a constraint on the amount of labor firms can hire. In particular, we assume that firms can hire no more than F workers. To simplify notation, we set F = 1. This constraint can be motivated as a limitation on the span of control by managers, as in Lucas (1978).14 Output produced by firm j in country i at time t, which we denote by Yj it , is given by Yj it = Kjαit (Aj it hi )1−α . (We now use time subscripts because they clarify the maximization problem below.) The firm can convert output into consumption, investment goods or R&D according to Yj it = Cj it + pi Ij it + Rj it , and the firm’s capital stock evolves according to K˙ j it = 14 If one takes F = 1 literally, then the externalities are in the human capital investment of individual workers.
844
P.J. Klenow and A. Rodríguez-Clare
Ij it − δKj it . Finally, the firm’s technology index Aj it evolves according to: A˙ j it = (1 − µ)λRj it + µλR it + εAj it 1 − Aj it /A∗it
(4.9)
where µ is a parameter between zero and one, R it is the average of Rj it across firms in country i (we use the bar over the variable to emphasize that this is the average across firms, and not the aggregate economy-wide variable), and A∗it is the technology frontier for country i with A˙ ∗it /A∗it = gA for all i in steady state. There are two features in this specification that merit some explanation. First, the “benefits of backwardness” are determined by the term 1 − Aj it /A∗it , which can differ across firms in country i: a more backward firm in country i would have a higher catchup term. If instead we specified the catch-up term as 1 − Ait /Ait ∗ (where Ait is the average technology index across firms in country i), then there would be a negative externality because, as a firm does more research, it increases the country’s average technology index and decreases the catch-up term for the other firms. Given that there is no particular reason to think that this negative externality is a relevant feature to include in the model, we have chosen to specify the catch-up term as 1 − Aj it /A∗it . Second, there is a positive research externality across firms within each country, represented by the term µλR it . This externality captures the idea that a firm benefits directly from research undertaken by other firms within the same economy. To relate this to what we had in subsection 4.1, note that if firms within a country are identical, then Rj it = R it and Aj it = Ait . Using this in (4.9), we obtain: ˙ = λR + εA 1 − A /A∗ . A it it it it it But note that Ait = Ait and R it = Rit /Lit , where Lit is the total labor force in country i and also the number of firms there, given our assumptions above. Using these results and noting that SRi = Rit /Yit we obtain Equation (4.2). Firms in country i pay taxes at rate τKi on capital income (output less the wage bill), and there is an R&D tax (or subsidy, if it is negative) of τRi .15 We stress that this R&D tax parameter does not have to be interpreted strictly as a formal tax or subsidy; when positive, the R&D tax parameter τRi could also be interpreted as capturing “barriers to technology adoption”, as in Parente and Prescott (1994).16 The firm’s dynamic optimization problem is to choose a path for Rj is and to maximize ∞ (1 − τKi )[Yj is − wis ] − pi Ij is − (1 + τRi )Rj is e−r(s−t) ds t
15 We should note here that the tax rate on capital income also affects the incentive to do research. The
notation used for the two tax rates is meant to emphasize that τKi affects all forms of accumulation by the firm, whereas τRi only affects research expenditures. 16 We assume that any tax revenue collected is distributed back to consumers in lump-sum fashion.
Ch. 11:
Externalities and Growth
845
subject to K˙ j is = Ij is − δKj is , A˙ is /Ais = A˙ ∗is /A∗is = gA , and A˙ j is = (1 − µ)λRj is + µλR is + εAj is 1 − Aj is /A∗is . As shown in Appendix A, by imposing the symmetry condition on the two Euler equations for this optimization problem, we obtain the following two conditions for the symmetric equilibrium: 1 − τKi pi Kit , =α Yit r +δ Ωi (1 − α)λki (1 − ai ) − gA ai /(1 − ai ) + ε(1 − ai ) = r
(4.10) (4.11)
where Ωi ≡
(1 − τKi )(1 − µ) . (1 + τRi )
Equation (4.10) defines the equilibrium capital-output ratio in country i and Equation (4.11) implicitly defines the equilibrium relative A in country i. Given ai and knowing ki from the data, we can plug their values into Equation (4.3) to obtain the equilibrium steady state R&D investment rate, sRi . It is easy to see that an increase in the capital income tax or the R&D tax or an increase in the externality parameter, µ, would decrease Ωi and hence lead to a decline in equilibrium ai (this is because the left-hand side of (4.11) is decreasing in ai ). This, of course, would imply a decline in the R&D investment rate. The same reasoning shows that ai is increasing in ki but it is not necessarily the case that sRi increases with ki (see Appendix A). Combining the result for the social rate of return in Equation (4.5) with (4.11), we obtain the following expression for the wedge between the social and private rate of return to R&D: r˜i − r = (1 − Ωi )(1 − α)λki (1 − ai ) + gL .
(4.12)
The first term on the right-hand side is the distortion created by Ω, which captures the effect of the income tax, τK , the R&D tax, τR , and the externality parameter, µ. If there are no taxes and µ = 0 (no domestic R&D externalities), then Ωi = 1 and the wedge between the social and private rate of return to R&D collapses to gL .17 4.4. Calibration The model described in the previous section, together with the (JEK) formulation for international spillovers with ηil = e−θd(i,l) , constitutes the model we calibrate in this subsection. Since we will only be working with the symmetric steady state equilibrium,
17 As explained above, g is associated with a positive externality because new firms start up with the same L
productivity as existing firms. Since the number of firms is equal to the workforce, the value of ideas and the social rate of return are increasing in gL .
846
P.J. Klenow and A. Rodríguez-Clare
in this subsection we suppress time and firm subscripts to simplify notation. Given N countries, the steady state equilibrium is given by {Ki , Yi , ki , ai , sRi , Ai , A∗i , zi ; i = 1, . . . , N } such that: gL gA = (4.13) , 1−γ 1 − τKi pi Ki , =α (4.14) Yi r +δ α/(1−α) Ki ki = hi (4.15) , Yi gA ai + ε(1 − ai ) = r, Ωi (1 − α)λki (1 − ai ) − (4.16) 1 − ai gA , ai = 1 − (4.17) λsRi ki + ε Ai = ai A∗i , (4.18) ∗ ˜ Ai = zi A, (4.19) A˜ = (vL)1/(1−γ ) , σ v= λkl sRl al ωl , gA l σ (1−γ )+1 zi = λkl sRl al ωl zl e−θd(i,l) vgA
(4.20) (4.21) (4.22)
l
where the last equation comes from (JEK) together with (4.8). If we knew the relevant parameters and tax rates and wanted to solve for an equilibrium, we would first start by solving for gA from Equation (4.13). Given data for exogenous variables hi , pi and τKi we could then calculate equilibrium ki using (4.14) and (4.15). Together with gA and parameter ε, Equation (4.16) would yield ai . From (4.17) we would then obtain sRi . Up to this point, there is no interaction across countries, so these results do not depend on geography or θ ; this dimension becomes relevant in obtaining actual productivity levels, because they depend on the variables zi , which capture spillovers from the rest of the world to country i. To see how this operates, note that given the value of θ , Equation (4.22) configures a system of N equations (where N is the number of countries) in N unknowns (z1 , z2 , . . . , zN ). The solution ˜ which to this system determines zi . Given parameter σ , Equation (4.20) determines A, ∗ together with zi determines each country’s technology frontier Ai (Equation (4.19)). Finally, from Equation (4.18), a country’s technology frontier together with its relative A level ai determines Ai . For the calibration exercise, the first step is to specify the variables we observe and ∗ how they relate to the model. We take human capital to be hi = eϕ MY Si , where MY Si is mean years of schooling of the adult population in country i, obtained from Barro and Lee (2000). We use R&D data from Lederman and Saenz (2003). The 48 countries in our sample are the ones for which there is R&D data for 1995, as well as the necessary
Ch. 11:
Externalities and Growth
847
TFP and capital intensity variables described in Section 3. The first two columns of Table 7 reproduce the values of the R&D investment rate and the value of A for the 48 countries in our sample. For the basic parameters we use the following values: ϕ = 0.085, α = 1/3, δ = 0.08, gL = 0.011 and gA = 0.015. For the first three, see our discussion in Section 3. The last two (the growth rates) were obtained from OECD average growth rates of L and A for the period 1960–2000.18 Using (4.13), the values for the two growth rates imply γ = 0.31. To calculate the net private rate of return, r, which we assume to be common across countries, we take the capital income tax in the U.S. to be 25% (τK,US = 0.25).19 Given the 1995 U.S. nominal capital–output ratio of 1.5 (see Section 3 for how we constructed capital–output ratios), this implies from (4.14) that r = 8.6%. Given this level for r, we then use Equation (4.14) together with country nominal capital–output ratios to obtain each country’s implicit income tax τKi . Remaining parameters we must calibrate are ε, λ, µ and θ . Unfortunately, there is no empirical work that we can rely on to pin down ε. Thus, we choose a value for ε based on the following reasoning. First, ε cannot be much higher than gA . This is because for ksR 0 Equation (4.17) implies that a 1−gA /ε. Thus, a high value of ε would imply that some countries’ relative empirical A becomes lower than the theoretical minimum 1 − gA /ε. In other words, if free technology diffusion is too important, then it would be hard to account for countries with very low A levels. Second, if ε < gA , then countries with a low value of ksR (λsR k < gA − ε) would not be able to keep up with the world’s rate of growth in technology, so they would not have a steady state relative A level. (Consistent with stable long run relative income levels, Figure 3 showed roughly parallel slopes for average income across deciles over 1960–2000, with each decile based on 1960 income.) Thus, it seems reasonable to impose the intermediate condition that ε = gA . We believe, however, that future empirical work should attempt to understand the importance of free technology diffusion captured by parameter ε. Given this choice for ε, we use two empirical findings to pin down parameters λ and µ, namely that the social rate of return to R&D in the U.S. is three times the net private rate of return [Griliches (1992)] and that the U.S. imposes a subsidy of 20% on R&D [Hall and Van Reenen (2000)], implying that τR,US = −0.2.20 Given data for sR and k for the U.S. in 1995 (sR,US = 2.5% and kUS = 3.6), then this restriction together with Equation (4.17) implies aUS = 0.7 and λ = 0.38.21 From (4.16) we then obtain µ = 0.55. 18 Specifically, the growth rate of A is the annual growth rate of the weighted average of A in the OECD with
weights given by employment levels in 1960. OECD membership is defined by 1975 status. 19 Auerbach (1996) estimates an effective tax rate in the U.S. of about 16%, but King and Fullerton (1984)
estimate a much higher level of around 35%. We use 25% as an intermediate value. 20 The 20% subsidy on R&D is the statutory rate, but not necessarily the effective one. Bronwyn Hall sug-
gested that a 5% subsidy would be a better approximation to the effective rate. We performed an alternative calibration with tR,US = −0.05 and the results did not change significantly. 21 Due to the non-linearity of the expression for the social rate of return to R&D, there are actually two values of λ which are compatible with a social rate of return equal to 26% (three times the private rate of return).
848
P.J. Klenow and A. Rodríguez-Clare Table 7 Data and “true” values for research intensity and productivity Country Argentina Bolivia Brazil Chile China Colombia Ecuador Egypt Hong Kong Hungary Indonesia India Israel South Korea Mexico Panama Peru Poland Romania Senegal Singapore El Salvador Thailand Tunisia Taiwan Uganda Uruguay Venezuela Austria Belgium Canada Denmark Spain Finland France United Kingdom Germany Greece Ireland Italy Japan Netherlands Norway New Zealand Portugal Sweden Turkey USA
Data sR
Data A
“True” sR
Implied A
0.41% 0.37% 0.86% 0.61% 0.60% 0.28% 0.08% 2.11% 0.25% 0.73% 0.09% 0.63% 2.75% 2.49% 0.31% 0.38% 0.05% 0.69% 0.80% 0.02% 1.16% 0.33% 0.12% 0.32% 1.78% 0.59% 0.28% 0.48% 1.56% 1.57% 1.64% 1.84% 0.81% 2.37% 2.31% 1.99% 2.25% 0.49% 1.35% 1.08% 2.89% 1.99% 1.71% 0.97% 0.57% 3.46% 0.38% 2.51%
9,720 4,672 9,836 11,078 2,570 8,143 5,990 11,126 17,874 7,172 5,912 3,755 13,919 8,842 8,781 6,106 4,285 4,893 2,757 3,069 13,592 11,096 5,212 10,323 14,944 2,878 10,088 9,427 14,807 15,597 11,614 13,678 15,758 10,358 15,411 13,954 11,993 10,046 17,177 19,204 9,864 14,136 10,990 9,911 13,230 10,416 7,800 15,472
1.21% 0.74% 1.67% 1.98% 0.28% 1.54% 0.69% 3.57% 5.49% 0.63% 0.91% 0.60% 2.15% 0.71% 1.08% 0.60% 0.40% 0.33% 0.16% 0.64% 2.16% 3.26% 0.49% 2.11% 3.59% 1.02% 1.69% 1.35% 2.60% 2.89% 1.12% 1.95% 3.69% 0.94% 3.07% 2.35% 1.31% 1.07% 5.08% 8.27% 0.85% 2.19% 0.88% 0.85% 2.65% 0.91% 1.18% 2.51%
9,719 4,672 9,835 11,075 2,570 8,141 5,990 11,119 17,732 7,172 5,911 3,755 13,922 8,843 8,780 6,106 4,285 4,893 2,757 3,068 13,587 11,084 5,212 10,319 14,928 2,878 10,085 9,426 14,800 15,586 11,615 13,677 15,726 10,360 15,404 13,952 11,994 10,046 17,098 18,795 9,865 14,135 10,991 9,911 13,220 10,418 7,800 15,472
Ch. 11:
Externalities and Growth
849 Table 8 Model A versus data A (θ = 0 case)
Country Quartile 1 Quartile 2 Quartile 3 Quartile 4
Data k
Data sR
Data A
Model A
2.0 2.5 3.1 2.9
0.4% 0.5% 1.7% 1.7%
4,478 9,574 11,111 15,441
2,184 5,358 11,763 12,286
A parameter remaining to calibrate is θ .22 Before discussing possible values for this parameter, it is useful to consider the case where θ = 0 – so that there is no effect of distance on international spillovers – and to compare the implications of the model to the data. Using the R&D investment rate data of Lederman and Saenz (2003) and our estimated k levels, Equation (4.17) yields the model’s implied relative A level for each country (ai ). We want to compare this against the data. To do so, we use the value of A we calculated for the U.S. in the previous section and aUS = 0.7 to obtain an implied value for the world technology frontier, A∗ (recall that with θ = 0 there is a well defined technology frontier that is common to all countries). We can then obtain the model’s implied A values for all countries using Ai = ai A∗ . The result of this exercise is shown in Table 8, where we divide countries into four groups according to their levels of A and show the median of the different variables for each group. It is clear that the model does badly for the poorest countries, predicting much lower A levels for them than occur in the data. This discrepancy does not occur for the richest countries, so the model is predicting significantly larger A differences than in the data. For example, whereas (according to the data) the top group’s median A is 3.4 times the median A of the bottom group, the model implies a ratio of 5.6. The model implies large differences in productivity in response to small differences in R&D investment rates. As is well known, the neoclassical model – with only around 1/3 share for physical capital – cannot generate large differences in steady state labor productivity in response to modest differences in investment rates [see the discussion in Lucas (1990)]. It is worth pausing here to explore some of the reasons behind these divergent properties. Manipulating the neoclassical model, one can show that the semielasticity of steady state labor productivity with respect to the investment rate is given The higher value of λ, however, would imply a high relative A level for the U.S. and consequently – given measured A for the U.S. – a value for A∗ that would be lower than the measured A levels of the high A countries, such as Hong Kong and Italy. To avoid this, we choose the lower value of λ. 22 We must also set a value for σ , which is crucial for determining the level of A. ˜ We use the value of AUS obtained from the data, (4.18)–(4.21), aUS = 0.7, and a value for zUS (equal to one when θ = 0 and a known value from the solution to the above system of equations for the case θ > 0) to arrive at a value for σ . In the case with θ = 0, then zUS = 1, and hence A˜ = AUS /aUS = 22,196. Given the value of aggregate L in our sample of countries, this implies from (4.20) that ν = 0.00082, which plugging into Equation (4.21) implies finally that σ = 0.0023.
850
P.J. Klenow and A. Rodríguez-Clare
by: r 1 ∂ ln y = · . ∂s 1 − α δ + gA + gL
(4.23)
With the values we used above (α = 1/3, δ = 0.08, gL = 0.011 and gA = 0.015), (4.23) yields a semi-elasticity of only 1.22% when evaluated at r = 8.6%. Thus large differences in investment rates would be required to generate sizable differences in labor productivity across countries. Two differences between the way the R&D investment rate operates in our model and the way the physical capital investment rate operates in the neoclassical model stand out: first, the depreciation rate of ideas in our model is zero versus δ = 0.08 for capital in the neoclassical model; second, the elasticity of output with respect to the stock of ideas can exceed 1/3 (we have it at 2/3). To see the importance of these values, note that with α = 2/3 the semi-elasticity doubles to 2.46% (still with r = 8.6%). If we use δ = 0 as well, then the semi-elasticity increases to 9.6%. In our model, the combined share of physical capital and ideas is actually 1. Without the constraint of the world technology frontier, therefore, the long run response of output would be infinite. It is important to recall that the results shown in Table 8 and discussed above were derived for the case of θ = 0. Is it possible that a positive value of θ could improve the model’s fit with the data? As will become clear below, countries with high levels of k and high R&D investment rates tend to cluster together. Thus, assuming a positive value for θ would actually make the model less consistent with the data, since it would imply an even larger difference between A levels across rich and poor countries. One possible reason why the model is not doing well in matching the data is that measured R&D is not the appropriate empirical counterpart of “research” in the type of models we have been examining. In particular, measured R&D only includes formal research; this is research performed in an R&D department of a corporation or other institution. This fails to capture informal research, which may be particularly important in non-OECD countries. To explore this idea, in the rest of this section we assume that both R&D intensity and the productivity index A are measured with error. We estimate “true” R&D intensities by minimizing a loss function equal to the sum of two terms that capture, respectively, the deviation of the “true” R&D intensities from the data and the deviation of the model’s implied (log of) A values from the data, with weights given by the standard deviation of the corresponding differences.23 In principle, we could follow this procedure for each value of θ . However, at θ = 0, the partial derivative of our loss function with respect to θ is positive and large, implying that – just as argued above – the model’s fit with the data worsens as θ increases from zero. Thus, we restrict ourselves
23 We do this in two stages. In the first stage we minimize a loss function without weights. We use the results
to calculate the standard deviation of the error terms, or differences between data and “true” values for both R&D intensity and productivity. In the second stage we minimize the loss function with weights given by these calculated standard deviations.
Ch. 11:
Externalities and Growth
851
Figure 4. Deviations of the model from the data for research intensity and productivity.
to estimating R&D intensities for θ = 0 and later show what happens if, keeping the same R&D intensities estimated for θ = 0, we have positive values of θ . It should be acknowledged that this procedure obviously implies that we can no longer evaluate the model’s consistency with the data; our interest is now to explore the implications of the model for the differences in R&D investment rates that would be necessary to explain cross-country differences in A, as well as the implied differences in R&D tax rates that would be necessary to bring about those R&D investment rates.24 The results of the exercise described above are shown in Figure 4 and Table 7 (columns 3 and 4). There are three points to note from these results. First, it is clear
24 This exercise assumes that all differences in A across countries are causes by differences in technology
adoption. An alternative view is that A differences arise from “distortions” that directly affect productivity through their impact on factor allocation, input usage, effort on the job, etc. We believe that understanding the relevance of barriers to technology adoption versus distortions is an important topic for future research.
852
P.J. Klenow and A. Rodríguez-Clare
that the procedure leads to only small deviations of A from the data, whereas the deviations are more significant for R&D intensities. It would appear that R&D intensities have more significant measurement problems (or are conceptually more different than research intensity in our model) than productivity levels. Indeed, the standard deviation of residuals of sR with respect to the data is 0.12, whereas the corresponding value for the (log of) A is 0.01.25 Second, there are some countries for which the estimated R&D intensity is much higher than the data. Italy, for example, has a measured R&D intensity of 1.1%, whereas its “true” value is 8.3%. This arises because of Italy’s high measured productivity (Italy’s A is 24% higher than the U.S. level) and low value of k (2.6 versus 3.6 in the U.S.). Something similar happens for other high-A countries, such as Hong Kong and Ireland. Finally, just as one would expect given the results above, estimated R&D intensities vary much less than the corresponding values in the data. This is the main mechanism by which the procedure allows the model to fit perfectly. It also suggests that measurement error may be behind the low R&D intensities of several poor countries and of some high A countries such as Italy, Ireland and Hong Kong. We can now explore what happens when θ is positive, so that spillovers decline with distance. Given the estimated R&D intensities, productivity levels change with θ only because of the associated changes in the variables z, which capture the effect of distance on spillovers for each country. In principle, we can obtain the values of (z1 , z2 , . . . , z48 ) for any θ 0 from the solution of a system of 48 non-linear equations represented by (4.22). Equation i of this system can be expressed as: 48 λkl sRl al zl ωl e−θd(i,l) (1−γ )+1 = l=1 48 . zi (4.24) l=1 λkl sRl al ωl Solving this system numerically for the parameter values we have discussed and the R&D intensities derived before, we arrive at a value of zi for each country, from which we can then obtain the country’s level of A by using Ai = ai zi A˜ from (4.18) and (4.19). What are reasonable values to use for the parameter θ ? Using industry level data on productivity and research spending across the G-5 countries, Keller (2002) estimated a reduced form model where cumulative industry research affects own productivity and also affects productivity in the same industry in other countries through international spillovers that decline with distance.26 Given the similarity between Keller’s system and a reduced form of our model, it seems reasonable to use Keller’s estimate of θ , namely θK ≡ 0.0009 in the calibration of our model. It turns out, however, that with θ = θK our model cannot match the data – in particular, there is no solution to the system of Equations (4.24), at least for the parameters used for the exercises above. 25 These standard deviations are the ones that arise after the two stage procedure described in the previous
footnote. After the first stage, the standard deviations for the R&D rate and the (log of) A are 0.11 and 0.03, respectively. 26 For other estimates of international spillovers from R&D, see Coe and Helpman (1995) and Coe, Helpman and Hoffmaister (1997). For a study of agricultural R&D spillovers, see Evenson and Gollin (2003). Becker, Philipson and Soares (2003) present evidence consistent with international spillovers of health technology.
Ch. 11:
Externalities and Growth
853 Table 9 Implied R&D tax rates “True”
Country Quartile 1 Quartile 2 Quartile 3 Quartile 4
τK
k
sR
a
SRR
13% 0% 4% 6%
2.0 2.5 3.1 2.9
0.60% 1.13% 1.97% 2.98%
20% 43% 50% 70%
42% 37% 29% 21%
τR 102% 93% 31% −16%
a for τRi = τR,US 58% 68% 72% 70%
Notes: τR is calculated as the level of τR needed to generate the “true” research intensity. For each country, we use its own implied income tax level (τK ) and its own capital intensity level k. The last column presents the equilibrium steady state relative A level (a) for the hypothetical case in which all countries have the same R&D tax as the U.S. (τRi = τR,US ) but have different income tax rates and capital intensity levels.
This is because θK is unreasonably high. One way to see this is by noting that it implies a half distance of 746 miles: this implies that spillovers from the U.S. to Japan would be only one tenth of those to Mexico, and spillovers from the U.S. to New Zealand would be only one fifth of those to Japan. We were able to find solutions for the system with θ = θK /5. For comparison, we also obtained solutions for two other values of θ , namely θ = θK /10 and θ = θK /100. A group of European countries (Belgium, France, United Kingdom, Germany, Ireland, Italy, and Netherlands) always come out with the highest values of z, whereas New Zealand always comes out with the lowest value. For θ = θK /100, θ = θK /10 and θ = θK /5, the minimum and maximum values of z are (93%, 96%), (48%, 68%) and (24%, 50%), respectively. Clearly, for high values of θ , geography by itself can lead to large differences in productivity across countries. In the rest of this section, we focus on the case θ = 0, since – as explained above – the model’s fit with the data is best at this point. (Recall that the model fits perfectly because we are using the estimated research intensities and the implied A values). Table 9 presents summary statistics for the solution for the case of θ = 0. Our discussion of these results will focus on the comparison of the poorest and richest quartiles (ordered, as above, in terms of A levels) in this table. There are several points that we want to highlight in relation to these results. First, the median income tax is 13% and 6% for the poorest and richest countries, respectively. Everything else equal, this would lead to a lower R&D investment rate in the poorest countries. Second, as expected, rich countries have a higher k than poor countries: the level of k in these two groups is 2 and 2.9, respectively. As commented in Section 4.2, higher k has a direct effect on relative A (see Equation (4.17)) and an indirect effect (it could be positive or negative) through its impact on R&D investment rates (see Equation (4.16)). A natural question arises: is it the case that once we take into account the effect of k on TFP we can resuscitate the “neoclassical revolution” mantra that differences in physical and human capital accumulation rates account for most of cross-country income differences? More concretely, how much of the variation in A
854
P.J. Klenow and A. Rodríguez-Clare
levels across countries is due to the variation in levels of k? A simple way to answer this question is to note from Equation (4.17) that differences in relative A levels are driven by differences in the product sR k across countries. Running a regression of sR on the log of this product yields a coefficient of 0.8, which implies that when sR k increases by one percent, we should expect sR to increase by 0.8%. Clearly, most of the variance of the product sR k is accounted for by the variance of sR . Third, the social return to R&D is higher for poor countries. This is consistent with the findings in Lederman and Maloney (2003) and also with the idea that poor countries have policies and institutions that negatively affect the quantity of research. Fourth, the column with heading τR indicates the R&D tax rate required to produce the “true” R&D investment rates given each country’s levels of τK . The main question we address here is whether differences in income tax rates, which affect both the rate of investment in physical capital and R&D, are sufficient to explain differences in estimated research intensities. The answer is clearly negative: the required R&D tax rate in the poorest countries is 102% compared to −16% in the richest countries. To address the same question from a different angle, the last column calculates each country’s implied relative A level if all countries had the same R&D tax as the U.S. but kept their own levels of τK . It is clear that differences in τK alone are too small to account for the wide dispersion in productivity levels across countries. Finally, as emphasized above, the results in Table 9 suggest that small differences in steady-state R&D investment rates have large effects on steady state relative A levels. For example, in the calibrated model, by increasing its R&D investment rate by 1% from 0.6% India could double its steady state relative A level from 17% to 34%, clearly a very large effect. India’s social rate of return to research, however, is a moderate 30%. The apparent contradiction arises because the large effect of the increase in the R&D intensity on the relative A level is a steady-state comparative-statics result, and hence does not take into account the transition, which is a crucial component in the calculation of the social rate of return to R&D. As a result, in spite of the large effect of differences in R&D investment rates on relative A levels in steady state, the required implicit taxes on R&D are not huge. 4.5. The benefits of engagement One of the benefits of the model we have constructed is that it allows us to perform an interesting exercise. We can ask: how much do countries benefit from spillovers from the rest of the world?27 First, note that a country’s equilibrium ai is not affected by being isolated or engaged. Thus, the whole benefit of engagement is going to captured by the way engagement affects the term zi . Now, if a country is isolated, or disengaged, its equilibrium z would be characterized by the solution to the system (4.24) when θ → ∞. It is easy to check 27 See Eaton and Kortum (1999) for a host of related counterfactuals with similar outcomes.
Ch. 11:
Externalities and Growth
855
Table 10 Benefits of engagement for selected countries Country
Share of world’s L
U.S. UK Belgium Brazil India China Senegal
7.1% 1.5% 0.2% 3.1% 1.3% 38.7% 0.2%
Scale effect 37 297 4,093 114 9 4 4,451
S.V. effect
Total effect
0.12 0.21 0.12 0.97 23.0 70.6 42.0
5 64 480 110 217 258 187,035
that this yields
1/(1−γ ) λki sRi ai ωi z˘i = . l λkl sRl al ωl
(4.25)
Thus, the benefits of engagement are captured by zi /˘zi . From (4.17) we get
1/(1−γ ) l ωl υl zi /˘zi = zi ωi υi where υi ≡ ai λki sRi = λRi /A∗i Li is a measure of research intensity. Letting υ¯ ≡ j ωj υj be the world’s weighted average of υi , we obtain 1/(1−γ ) 1/(1−γ ) zi 1 υ¯ = zi . z˘i ωi υi
(4.26)
The first term on the RHS of this equation, zi , captures the fact that even when fully engaged, a country’s technology frontier is inferior to the world’s frictionless frontier if θ > 0, in which case zi < 1 for all i. The second term is the pure scale effect that arises in this model. The third term, which we call the “Silicon Valley” effect, captures the fact that richer countries benefit less from being part of the world than poor countries do because of their higher effective research intensity. Table 10 presents results based on these values and assuming θ = 0, which implies zi = 1 for all i. The results suggest huge benefits of engagement. At the extreme, Senegal’s productivity is 187 thousand times higher than it would be if it was isolated! Of course, if θ > 0 then zi < 1 and the overall effect would be small. Still, it is our conjecture that any reasonable value of θ would still imply enormous benefits of engagement. Of course, in a more general model, it is reasonable to think that productivity could not fall below a certain level because of Malthusian forces. Specifically, suppose there is a fixed factor such as land. Then, for sufficiently low A, population would decline until income per capita was equal to the subsistence level. Instead of very low levels of A, disengagement would mean very low population sizes. Put differently, an important part of the benefits of engagement may be realized through larger population rather than
856
P.J. Klenow and A. Rodríguez-Clare
higher productivity. The implications are clear: if it were not for the benefits of sharing knowledge internationally, countries would have much lower productivity levels and populations than they now do. 4.6. Discussion of main results We finish this section with a discussion of the main results we want to emphasize. First, the usual separation between capital and productivity – or between investment and technological change – is not always valid. For a given R&D investment rate, higher investment rates in physical and human capital lead naturally to higher TFP productivity levels. Thus, one should not jump from cross-country dispersion in TFP to the conclusion that differences in physical and human capital play a minority role in accounting for international income differences. When we calibrate our model, however, we find that differences in R&D investment rates account for most of the cross country variation in productivity. Second, international variation in R&D investment appears more than large enough to generate the international variation in productivity. But it seems likely that measured R&D does not capture all of the investment associated with adoption of foreign technology. Indeed, we find that countries such as Indonesia, Peru and Senegal have R&D investment rates that are much too low to be consistent with their productivity levels. It is likely that their true research intensities are much higher than the measured ones. We hope to see more research in understanding how to capture and measure “research”. Third, differences in (implicit) capital income tax rates are not large enough to account for the observed differences in R&D investment rates and productivity levels. The calibrated model suggests that sizable differences in R&D taxes are needed. These R&D taxes are clearly not formal or explicit taxes, but the result of policies and institutions that make research more costly or reduce its associated returns. Exploring the nature and source of these differences in implicit R&D taxes across countries is an important topic for future research. Finally and most importantly, the calibrated model indicates that countries benefit enormously from international knowledge spillovers. We think any reasonable value of θ (which governs the rate at which spillovers decline with distance) would yield results similar to those we presented above.
5. Conclusion Externalities are not theoretically necessary to sustain growth. But they appear essential for understanding why many countries grow at similar rates despite differing investment rates. A dramatic way to summarize the importance of international knowledge externalities is to calculate world GDP in the absence of such externalities. According to our calibrated model, world GDP would be only 6% of its current level, or on the order of $3 trillion rather than $50 trillion, if countries did not share ideas. Such scale effects
Ch. 11:
Externalities and Growth
857
from the nonrivalry of knowledge are a central theme in the works of Romer (1990), Kremer (1993), Diamond (1997), Jones (2001, 2005) and many others. Because diffusion is not costless, however, differences in knowledge investments may explain a significant portion of income differences across countries. We show that modest barriers to technology adoption could account for differences in TFP of a factor of four or more, as observed in the data. But we have not documented such barriers to knowledge diffusion in practice. We consider this a priority for future research. A related issue is that the high power of the model in explaining large productivity differences from small barriers to technology adoption comes at a cost, namely that the transition from one steady state to another after a discrete change in barriers takes a long time. This is what allows small changes in barriers to have large steady-state effects without implying huge social rates of return to research in poor countries. But it implies that the model is not very useful in understanding miracles. Perhaps a different formulation of the “benefits of backwardness” in Equation (4.2) would allow for a more attractive balancing of steady state and transition properties. We have also left for future research the identification of the primary channels of international knowledge spillovers. Trade, joint ventures, FDI, migration of key personnel, and imitation may all play important roles. See Keller (2004) for a survey of recent empirical work on this topic. A model with trade would lead naturally to some countries having a comparative advantage in doing innovative R&D and other countries focusing on adoption and imitation R&D. The evidence on international patenting supports the notion that innovative R&D is concentrated in rich countries. Of course, countries can imitate other imitators as well as the original innovators. We hope to see future research documenting not only the vehicles for knowledge diffusion, but their specific routes.
Acknowledgements For helpful comments we are grateful to Bob Hall, Chad Jones, Robert Lucas, Ben Malin, Romans Pancs, and Paul Romer.
Appendix A The firm’s maximization problem can be restated as choosing A˙ j is and K˙ j is to maximize: ∞ (1 − τKi )Kjαis (Aj is hi )1−α − pi K˙ j is − pi δKj is t
A˙ j is (1 + τRi ) − − εAj is − µλR is e−r(s−t) ds. (1 − µ)λ 1 − Aj is /A∗is
Letting Q represent the expression in the integral, then we know that a solution to this d (∂Q/∂ K˙ j is ) and problem must satisfy the following Euler equations: ∂Q/∂Kj is = ds
858
P.J. Klenow and A. Rodríguez-Clare
∂Q/∂Aj is =
d ˙ ds (∂Q/∂ Aj is ).
The first Euler equation is:
αYj is r +δ = . pi Kj is 1 − τKi Since in a symmetric equilibrium the capital–output ratio of firm j is the same as the aggregate capital output ratio, then this implies that: pi Kit 1 − τKi . =α Yit r +δ As to the second Euler equation, differentiation yields (we are using the symmetry condition for the equilibrium):
∂Q gA ai (1 + τKi ) = (1 − τKi )(1 − α)Yj is /Aj is − − ε e−r(s−t) ∂Aj is (1 − µ)λ (1 − ai )2 and ∂Q (1 + τRi ) e−r(s−t) . =− (1 − µ)λ(1 − ai ) ∂ A˙ j is Hence,
r(1 + τRi ) ∂Q d = e−r(s−t) . ds ∂ A˙ j is (1 − µ)λ(1 − ai )
Thus, the Euler equation is:
(1 − τKi )(1 − α)Yj is gA a i (1 + τRi ) r(1 + τRi ) . − − ε = Aj is (1 − µ)λ (1 − ai )2 (1 − µ)λ(1 − ai ) Noting that in a symmetric equilibrium we must have Yj is /Aj is = Yis /Ais Lis = ki , and manipulating, we get: Ωi (1 − α)λki (1 − ai ) −
gA ai + ε(1 − ai ) = r 1 − ai
where Ωi ≡
(1 − τKi )(1 − µ) . (1 + τRi )
Comparative statics (From here onwards we drop the subscripts.) It is easy to show that a is increasing in both Ω and k. In particular: ∂a Ω(1 − α)λ(1 − a) > 0. = ∂k Ω(1 − α)λk + (gA /(1 − a)2 ) + ε
Ch. 11:
Externalities and Growth
859
Differentiating gA = (λks + ε)(1 − a) (using s for sR ) we get (λs dk + λk ds)(1 − a) − da(λks + ε) = 0. This implies that: λk
∂s (∂a/∂k)(λks + ε) = − λs. ∂k (1 − a)
Plugging in from the result above we finally get: k
∂s Ω(1 − α)(λks + ε) = − s. ∂k Ω(1 − α)λk + (gA /(1 − a)2 ) + ε
Summing on the RHS and noting that the denominator is clearly positive we get that ∂s/∂k > 0 if and only if: gA Ω(1 − α)ε − s − εs. (1 − a)2 This term could well be negative.
References Acemoglu, D., Ventura, J. (2002). “The world income distribution”. Quarterly Journal of Economics 117, 659–694. Aghion, P., Howitt, P. (1992). “A model of growth through creative destruction”. Econometrica 60, 323–351. Aghion, P., Howitt, P. (1998). Endogenous Growth Theory. MIT Press, Cambridge, MA. Auerbach, A.J. (1996). “Tax reform, capital allocation, efficiency and growth”. In: Aaron, H.J., Gale, W.G. (Eds.), Economic Effects of Fundamental Tax Reform. Brookings Institution Press, Washington, DC, pp. 29–73. Barro, R.J., Lee, J.W. (2000). “International data on educational attainment: Updates and implications”. National Bureau of Economic Research Working Paper 7911. Cambridge, MA. Barro, R.J., Sala-i-Martin, X. (1995). Economic Growth. McGraw-Hill, New York. Baumol, W.J. (1986). “Productivity growth, convergence and welfare: What the long-run data show”. American Economic Review 76, 1072–1085. Becker, G.S., Philipson, T.J., Soares, R.R. (2003). “The quantity and quality of life and the evolution of world inequality”. National Bureau of Economic Research Working Paper 9765. Cambridge, MA. Bils, M., Klenow, P.J. (2000). “Does schooling cause growth?”. American Economic Review 90, 1160–1183. Caves, R.E. (1996). Multinational Enterprise and Economic Analysis, second ed. Cambridge University Press, New York. Coe, D.T., Helpman, E. (1995). “International R&D spillovers”. European Economic Review 39, 859–887. Coe, D.T., Helpman, E., Hoffmaister, A.W. (1997). “North–South R&D spillovers”. Economic Journal 107, 134–139. DeLong, J.B. (1988). “Productivity growth, convergence and welfare: Comment”. American Economic Review 78, 1138–1154. Diamond, J. (1997). Guns, Germs, and Steel: The Fates of Human Societies. Norton, New York. Easterly, W. (2001a). The Elusive Quest for Growth: Economists’ Adventures and Misadventures in the Tropics. MIT Press, Cambridge, MA. Easterly, W. (2001b). “The lost decades: Explaining developing countries’ stagnation in spite of policy reform 1980–1998”. Journal of Economic Growth 6, 135–157.
860
P.J. Klenow and A. Rodríguez-Clare
Easterly, W., Levine, R. (2001). “It’s not factor accumulation: Stylized facts and growth models”. World Bank Economic Review 15, 177–219. Easterly, W., Kremer, M., Pritchett, L., Summers, L.H. (1993). “Good policy or good luck? Country growth performance and temporary shocks”. Journal of Monetary Economics 32, 459–484. Eaton, J., Kortum, S. (1996). “Trade in ideas: Patenting and productivity in the OECD”. Journal of International Economics 40, 251–278. Eaton, J., Kortum, S. (1999). “International technology diffusion: Theory and measurement”. International Economic Review 40, 537–570. Evenson, R.E., Gollin, D. (2003). “Assessing the impact of the green revolution, 1960–2000”. Science 300, 758–762. Fischer, S. (1988). “Symposium on the slowdown in productivity growth”. Journal of Economic Perspectives 2, 3–7. Gollin, D. (2002). “Getting income shares right”. Journal of Political Economy 110, 458–474. Griliches, Z. (1992). “The search for R&D spillovers”. Scandinavian Journal of Economics 94, S29–S47. Grossman, G.M., Helpman, E. (1991). Innovation and Growth in the Global Economy. MIT Press, Cambridge. Hall, B.H. (1996). “The private and social returns to research and development”. In: Smith, B.L.R., Barfield, C. (Eds.), Technology R&D and the Economy. Brookings Institution, Washington, DC, pp. 140–183. Hall, B.H., Van Reenen, J. (2000). “How effective are fiscal incentives for R&D? A review of the evidence”. Research Policy 29, 449–469. Hall, R.E., Jones, C.I. (1999). “Why do some countries produce so much more output per worker than others?”. Quarterly Journal of Economics 114, 83–116. Hendricks, L. (2002). “How important is human capital for development? Evidence from immigrant earnings”. American Economic Review 92, 198–219. Heston, A., Summers, R., Aten, B. (2002). “Penn World Table Version 6.1”. Center for International Comparisons at the University of Pennsylvania (CICUP), Philadelphia, PA. Howitt, P. (1999). “Steady endogenous growth with population and R&D inputs growing”. Journal of Political Economy 107, 715–730. Howitt, P. (2000). “Endogenous growth and cross-country income differences”. American Economic Review 90, 829–846. Hummels, D., Klenow, P.J. (2005). “The variety and quality of a nation’s exports”. American Economic Review 95, 704–723. Jones, C.I. (1995). “R&D-based models of economic growth”. Journal of Political Economy 103, 759–784. Jones, C.I. (2001). “Was an industrial revolution inevitable? Economic growth over the very long run”. Advances in Macroeconomics 1, 1–43. Jones, C.I. (2002). “Sources of U.S. economic growth in a world of ideas”. American Economic Review 92, 220–239. Jones, C.I. (2005). “Growth and ideas”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1A. Elsevier, Amsterdam. This volume, Chapter 10. Jones, C.I., Williams, J.C. (1998). “Measuring the social return to R&D”. Quarterly Journal of Economics 113, 1119–1135. Jones, L., Manuelli, R. (1990). “A convex model of equilibrium growth: Theory and policy implications”. Journal of Political Economy 98, 1008–1038. Keller, W. (2002). “Geographic localization of international technology diffusion”. American Economic Review 92, 120–142. Keller, W. (2004). “International technology diffusion”. Journal of Economic Literature 42, 752–782. King, M.A., Fullerton, D. (1984). The Taxation of Income from Capital: A Comparative Study of the U.S., UK, Sweden and West Germany. University of Chicago Press, Chicago. Klenow, P.J., Rodríguez-Clare, A. (1997). “The neoclassical revival in growth economics: has it gone too far?”. In: Bernanke, B., Rotemberg, J. (Eds.), NBER Macroeconomics Annual. MIT Press, Cambridge, MA, pp. 73–103. Kortum, S. (1997). “Research, patenting, and technological change”. Econometrica 65, 1389–1419.
Ch. 11:
Externalities and Growth
861
Kremer, M. (1993). “Population growth and technological change: one million B.C. to 1990”. Quarterly Journal of Economics 108, 681–716. Kremer, M., Onatski, A., Stock, J. (2001). “Searching for prosperity”. Carnegie–Rochester Conference Series on Public Policy 55, 275–303. Lederman, D., Maloney, W.F. (2003). “R&D and development”. Unpublished paper. Office of the Chief Economist for LCR, The World Bank, Washington, DC. Lederman, D., Saenz, L. (2003). “Innovation around the world: a cross-country data base of innovation indicators”. Unpublished paper. Office of the Chief Economist for LCR, The World Bank, Washington, DC. Lucas, R.E. (1978). “On the size distribution of business firms”. Bell Journal of Economics, 508–523. Lucas, R.E. (1988). “On the mechanics of economic development”. Journal of Monetary Economics 22, 3–42. Lucas, R.E. (1990). “Why doesn’t capital flow from rich to poor countries?” American Economic Review 80, 92–96. Lucas, R.E. (2002). Lectures on Economic Growth. Harvard University Press, Cambridge. Lucas, R.E. (2004). “Life earnings and rural–urban migration”. Journal of Political Economy 112, S29–S59. Mankiw, N.G., Romer, D., Weil, D.N. (1992). “A contribution to the empirics of economic growth”. Quarterly Journal of Economics 107, 407–437. Parente, S.L., Prescott, E.C. (1994). “Barriers to technology adoption and development”. Journal of Political Economy 102, 298–321. Parente, S.L., Prescott, E.C. (2005). “A unified theory of the evolution of international income levels”. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of Economic Growth, vol. 1A. Elsevier, Amsterdam. This volume, Chapter 21. Pritchett, L. (1997). “Divergence, big time”. Journal of Economic Perspectives 11 (Summer), 3–17. Pritchett, L. (2005). “Does learning to add up add up? The returns to schooling in aggregate data”. In: Handbook of Education Economics. In press. Psacharopoulos, G., Patrinos, H.A. (2002). “Returns to investment in education: A further update”. World Bank Policy Research Working Paper 2881. Rebelo, S. (1991). “Long run policy analysis and long run growth”. Journal of Political Economy 99, 500– 521. Rivera-Batiz, L.A., Romer, P.M. (1991). “Economic integration and endogenous growth”. Quarterly Journal of Economics 106, 531–555. Romer, P.M. (1986). “Increasing returns and long-run growth”. Journal of Political Economy 94, 1002–1037. Romer, P.M. (1990). “Endogenous technological change”. Journal of Political Economy 98, S71–S102. Romer, P.M. (1994). “New goods, old theory, and the welfare costs of trade restrictions”. Journal of Development Economics 43, 5–38. Solow, R.M. (1956). “A contribution to the theory of economic growth”. Quarterly Journal of Economics 70, 65–94. Stokey, N.L. (1988). “Learning by doing and the introduction of new goods”. Journal of Political Economy 96, 701–717. Stokey, N.L. (1991). “Human capital, product quality, and growth”. Quarterly Journal of Economics 106, 587–617. Tamura, R.F. (1991). “Income convergence in an endogenous growth model”. Journal of Political Economy 99, 522–540. Young, A. (1998). “Growth without scale effects”. Journal of Political Economy 106, 41–63.
Chapter 12
FINANCE AND GROWTH: THEORY AND EVIDENCE ROSS LEVINE Department of Economics, Brown University and the NBER, 64 Waterman Street, Providence, RI 02912, USA e-mail:
[email protected]
Contents Abstract Keywords 1. Introduction 2. Financial development and economic growth: Theory 2.1. What is financial development? 2.2. Producing information and allocating capital 2.3. Monitoring firms and exerting corporate governance 2.4. Risk amelioration 2.5. Pooling of savings 2.6. Easing exchange 2.7. The theoretical case for a bank-based system 2.8. The theoretical case for a market-based system 2.9. Countervailing views to bank-based vs. market-based debate 2.10. Finance, income distribution, and poverty
3. Evidence on finance and growth 3.1. Cross-country studies of finance and growth 3.1.1. Goldsmith, the question, and the problems 3.1.2. More countries, more controls, and predictability 3.1.3. Adding stock markets to cross-country studies of growth 3.1.4. Using instrumental variables in cross-country studies of growth 3.2. Panel, time-series, and case-studies of finance and growth 3.2.1. The dynamic panel methodology 3.2.2. Dynamic panel results on financial intermediation and growth 3.2.3. Dynamic panel results and stock market and bank development 3.2.4. Time series studies 3.2.5. Novel case-studies 3.3. Industry and firm level studies of finance and growth 3.3.1. Industry level analyses
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01012-9
866 866 867 869 869 870 872 875 879 880 881 883 886 887 888 889 889 890 893 897 899 900 901 904 905 907 910 910
866
R. Levine 3.3.2. Firm level analyses of finance and growth 3.4. Are bank- or market-based systems better? Evidence 3.5. Finance, income distribution, and poverty alleviation: evidence
4. Conclusions Acknowledgements References
914 918 920 921 923 923
Abstract This paper reviews, appraises, and critiques theoretical and empirical research on the connections between the operation of the financial system and economic growth. While subject to ample qualifications and countervailing views, the preponderance of evidence suggests that both financial intermediaries and markets matter for growth and that reverse causality alone is not driving this relationship. Furthermore, theory and evidence imply that better developed financial systems ease external financing constraints facing firms, which illuminates one mechanism through which financial development influences economic growth. The paper highlights many areas needing additional research.
Keywords financial markets, economic development, financial institutions, technological change, corporate finance JEL classification: G0, O0
Ch. 12:
Finance and Growth: Theory and Evidence
867
1. Introduction Economists disagree sharply about the role of the financial sector in economic growth. Finance is not even discussed in a collection of essays by the “pioneers of development economics” [Meier and Seers (1984)], including three Nobel Prize winners, and Nobel Laureate Robert Lucas (1988, p. 6) dismisses finance as an “over-stressed” determinant of economic growth. Joan Robinson (1952, p. 86) famously argued that “where enterprise leads finance follows”. From this perspective, finance does not cause growth; finance responds to changing demands from the “real sector”. At the other extreme, Nobel Laureate Merton Miller (1998, p. 14) argues that, “[the idea] that financial markets contribute to economic growth is a proposition too obvious for serious discussion”. Drawing a more restrained conclusion, Bagehot (1873), Schumpeter (1912), Gurley and Shaw (1955), Goldsmith (1969), and McKinnon (1973) reject the idea that the financegrowth nexus can be safely ignored without substantially limiting our understanding of economic growth. Research that clarifies our understanding of the role of finance in economic growth will have policy implications and shape future policy-oriented research. Information about the impact of finance on economic growth will influence the priority that policy makers and advisors attach to reforming financial sector policies. Furthermore, convincing evidence that the financial system influences long-run economic growth will advertise the urgent need for research on the political, legal, regulatory, and policy determinants of financial development. In contrast, if a sufficiently abundant quantity of research indicates that the operation of the financial sector merely responds to economic development, then this will almost certainly mitigate the intensity of research on the determinants and evolution of financial systems. To assess the current state of knowledge on the finance-growth nexus, Section 2 describes and appraises theoretical research on the connections between the operation of the financial sector and economic growth. Theoretical models show that financial instruments, markets, and institutions may arise to mitigate the effects of information and transaction costs. In emerging to ameliorate market frictions, financial arrangements change the incentives and constraints facing economic agents. Thus, financial systems may influence saving rates, investment decisions, technological innovation, and hence long-run growth rates. A comparatively less well-developed theoretical literature examines the dynamic interactions between finance and growth by developing models where the financial system influences growth, and growth transforms the operation of the financial system. Furthermore, an extensive theoretical literature debates the relative merits of different types of financial systems. Some models stress the advantages of bank-based financial systems, while others highlight the benefits of financial systems that rely more on securities markets. Finally, some new theoretical models focus on the interactions between finance, aggregate growth, income distribution, and poverty alleviation. In all of these models, the financial sector provides real services: it ameliorates information and transactions costs. Thus, these models lift the veil that sometimes rises between the so-called real and financial sectors.
868
R. Levine
Section 3 reviews and critiques the burgeoning empirical literature on finance and growth, which includes broad cross-country growth regressions, times-series analyses, panel techniques, detailed country studies, and a recent movement that uses more microeconomic-based methodologies to explore the mechanisms linking finance and growth. Besides reviewing the results, I critique the empirical methods and the measures of financial development. Each of the different econometric methodologies that has been used to study the finance-growth nexus has serious shortcomings. Moreover, the empirical proxies for “financial development” frequently do not measure very accurately the concepts emerging from theoretical models. We are far from definitive answers to the questions: Does finance cause growth, and if it does, how? Without ignoring the weaknesses of existing work and the absence of complete unanimity of results, three tentative observations emerge. Taken as a whole, the bulk of existing research suggests that (1) countries with better functioning banks and markets grow faster, but the degree to which a country is bank-based or market-based does not matter much, (2) simultaneity bias does not seem to drive these conclusions, and (3) better functioning financial systems ease the external financing constraints that impede firm and industrial expansion, suggesting that this is one mechanism through which financial development matters for growth. I use the concluding section, Section 4, to (1) emphasize areas needing additional research and (2) mention the fast-growing literature on the determinants of financial development. In particular, this literature is motivated by the following question: If finance is important for growth, why do some countries have growth-promoting financial systems while others do not? Addressing this question is as fascinating and important, as it is multi-disciplined and complex. Developing a sound understanding of the determinants of financial development will require synthesizing and extending insights from many sub-specialties of economics as well as from political science, legal scholarship, and history. Before continuing, I want to acknowledge and emphasize that this review treats only cursorily some important issues. Here I highlight two, though this list is by no means exhaustive. First, I do not discuss in much depth the relationship between growth and international finance, such as cross-border capital flows and the importation of financial services. A serious discussion of international finance and growth would virtually double the length of this already long review. There is a critical theoretical, empirical, and policy question, therefore, that receives only limited attention in this essay: Can countries simply import financial services, or are there substantive growth benefits from countries having well-developed domestic financial systems? Second, I treat the political, legal, regulatory, and other policy determinants of financial development in only a perfunctory manner. This is a problem. The links between the functioning of the financial system and economic growth motivate research into the legal, regulatory, and policy determinants of financial development. Moreover, since the financial system influences who gets to use society’s savings, political forces have everywhere and always shaped financial sector policies and the operation of the financial system. Again, however, these crucial themes are beyond the scope of this paper. Instead, this chapter reviews the role
Ch. 12:
Finance and Growth: Theory and Evidence
869
of the financial system in economic growth and very briefly lists some ongoing work on the determinants of financial development in the conclusion.
2. Financial development and economic growth: Theory 2.1. What is financial development? The costs of acquiring information, enforcing contracts, and making transactions create incentives for the emergence of particular types of financial contracts, markets and intermediaries. Different types and combinations of information, enforcement, and transaction costs in conjunction with different legal, regulatory, and tax systems have motivated distinct financial contracts, markets, and intermediaries across countries and throughout history. In arising to ameliorate market frictions, financial systems naturally influence the allocation of resources across space and time [Merton and Bodie (1995, p. 12)]. For instance, the emergence of banks that improve the acquisition of information about firms and managers will undoubtedly alter the allocation of credit. Similarly, financial contracts that make investors more confident that firms will pay them back will likely influence how people allocate their savings. As a final example, the development of liquid stock and bond markets means that people who are reluctant to relinquish control over their savings for extended periods can trade claims to multiyear projects on an hourly basis. This may profoundly change how much and where people save. This section’s goal is to describe models where market frictions motivate the emergence of distinct financial arrangements and how the resultant financial contracts, markets, and intermediaries alter incentives and constraints in ways that may influence economic growth. To organize a review of how financial systems influence savings and investment decisions and hence growth, I focus on five broad functions provided by the financial system in emerging to ease information, enforcement, and transactions costs. While there are other ways to classify the functions provided by the financial system [Merton (1992), Merton and Bodie (1995, 2004)], I believe that the following five categories are helpful in organizing a review of the theoretical literature and tying this literature to the history of economic thought on finance and growth. In particular, financial systems: • Produce information ex ante about possible investments and allocate capital. • Monitor investments and exert corporate governance after providing finance. • Facilitate the trading, diversification, and management of risk. • Mobilize and pool savings. • Ease the exchange of goods and services. While all financial systems provide these financial functions, there are large differences in how well financial systems provide these functions. Financial development occurs when financial instruments, markets, and intermediaries ameliorate – though do not necessarily eliminate – the effects of information,
870
R. Levine
enforcement, and transactions costs and therefore do a correspondingly better job at providing the five financial functions. Thus, financial development involves improvements in the (i) production of ex ante information about possible investments, (ii) monitoring of investments and implementation of corporate governance, (iii) trading, diversification, and management of risk, (iv) mobilization and pooling of savings, and (v) exchange of goods and services. Each of these financial functions may influence savings and investment decisions and hence economic growth. Since many market frictions exist and since laws, regulations, and policies differ markedly across economies and over time, improvements along any single dimension may have different implications for resource allocation and welfare depending on the other frictions at play in the economy. In terms of integrating the links between finance and growth theory, two general points are worth stressing from the onset. First, a large growth accounting literature suggests that physical capital accumulation per se does not account for much of longrun economic growth [Jorgenson (1995, 2005)].1 Thus, if finance is to explain economic growth, we need theories that describe how financial development influences resource allocation decisions in ways that foster productivity growth and not aim the analytical spotlight too narrowly on aggregate savings. Second, there are two general ambiguities between economic growth and the emergence of financial arrangements that improve resource allocation and reduce risk. Specifically, higher returns ambiguously affect saving rates due to well-known income and substitutions effects. Similarly, lower risk also ambiguously affects savings rates [Levhari and Srinivasan (1969)]. Thus, financial arrangements that improve resource allocation and lower risk may lower saving rates. In a growth model with physical capital externalities, therefore, financial development could retard economic growth and lower welfare if the drop in savings and the externality combine to produce a sufficiently large effect. These ambiguities are general features of virtually all the models discussed below so I do not discuss them when describing each model. The remainder of this section describes how market frictions motivate the emergence of financial systems that provide five broad categories of financial functions and also describes how the provision of these functions may influence resource allocation and economic growth. 2.2. Producing information and allocating capital There are large costs associated with evaluating firms, managers, and market conditions before making investment decisions. Individual savers may not have the ability to collect, process, and produce information on possible investments. Since savers will be reluctant to invest in activities about which there is little reliable information, high information costs may keep capital from flowing to its highest value use. Thus, while many models assume that capital flows toward the most profitable firms, this presupposes that
1 For additional cross-country information, see King and Levine (1994) and Easterly and Levine (2001).
Ch. 12:
Finance and Growth: Theory and Evidence
871
investors have good information about firms, managers, and market conditions [Bagehot (1873, p. 53)]. Financial intermediaries may reduce the costs of acquiring and processing information and thereby improve resource allocation [Boyd and Prescott (1986)]. Without intermediaries, each investor would face the large fixed cost associated with evaluating firms, managers, and economic conditions. Consequently, groups of individuals may form financial intermediaries that undertake the costly process of researching investment possibilities for others. In Boyd and Prescott (1986), financial intermediaries look like banks in that they accept deposits and make loans. Allen (1990), Bhattacharya and Pfleiderer (1985), and Ramakrishnan and Thakor (1984) also develop models where financial intermediaries arise to produce information on firms and sell this information to savers. Unlike in Boyd and Prescott (1986), however, the intermediary does not necessarily both mobilize savings and invest those funds in firms using debt contracts. For our purposes, the critical issue is that financial intermediaries – by economizing on information acquisition costs – improve the ex ante assessment of investment opportunities with positive ramifications on resource allocation. By improving information on firms, managers, and economic conditions, financial intermediaries can accelerate economic growth. Assuming that many entrepreneurs solicit capital and that capital is scarce, financial intermediaries that produce better information on firms will thereby fund more promising firms and induce a more efficient allocation of capital [Greenwood and Jovanovic (1990)]. The Greenwood and Jovanovic (1990) paper is particularly novel because it formally models the dynamic interactions between finance and growth. Financial intermediaries produce better information, improve resource allocation, and foster growth. There is a cost to joining financial intermediaries, however. Growth means that more individuals can afford to join financial intermediaries, which improves the ability of financial intermediaries to produce better information with positive ramifications on growth. Thus, this research emphasizes (i) the two-way interactions between finance and growth and (ii) the relationship between income distribution and financial development during the process of economic development. Besides identifying the best production technologies, financial intermediaries may also boost the rate of technological innovation by identifying those entrepreneurs with the best chances of successfully initiating new goods and production processes [King and Levine (1993b), Galetovic (1996), Blackburn and Hung (1998), Morales (2003), Acemoglu, Aghion and Zilibotti (2003)].2 This lies at the core of Joseph Schumpeter’s (1912, p. 74) view of finance in the process of economic development:
2 Note, the model by Acemoglu, Aghion and Zilibotti (2003) focuses on examining when firms undertake innovative activities and when they adopt existing technologies from the world frontier. But, the existence of financial market frictions leads to financing constraints that help shape firm decisions. By implication, financial development will loosen those constraints and thereby affect innovative and adoption activities, with potential ramifications on aggregate growth rates.
872
R. Levine
The banker, therefore, is not so much primarily a middleman . . . He authorizes people in the name of society . . . (to innovate). Stock markets may also stimulate the production of information about firms. As markets become larger and more liquid, agents may have greater incentives to expend resources in researching firms because it is easier to profit from this information by trading in big and liquid markets [Grossman and Stiglitz (1980)] and more liquid [Kyle (1984) and Holmstrom and Tirole (1993)]. Intuitively, with larger and more liquid markets, it is easier for an agent who has acquired information to disguise this private information and make money by trading in the market. Thus, larger more liquid markets will boost incentives to produce this valuable information with positive implications for capital allocation [Merton (1987)]. Morck, Yeung and Yu (2000) provide tests of the information content of stock markets. While some models hint at the links between efficient markets, information, and steady-state growth [Aghion and Howitt (1998)], existing theories do not draw the connection between market liquidity, information production, and economic growth very tightly. Finally, capital market imperfections can also influence growth by impeding investment in human capital [Galor and Zeira (1993)]. In the presence of indivisibilities in human capital investment and imperfect capital markets, the initial distribution of wealth will influence who can gains the resources to undertake human capital augmenting investments. This implies a suboptimal allocation of resources with potential implications on aggregate output both in the short and the long run. 2.3. Monitoring firms and exerting corporate governance Corporate governance is central to understanding economic growth in general and the role of financial factors in particular. The degree to which the providers of capital to a firm can effectively monitor and influence how firms use that capital has ramifications on both savings and allocation decisions.3 To the extent that shareholders and creditors effectively monitor firms and induce managers to maximize firm value, this will improve the efficiency with which firms allocate resources and make savers more willing to finance production and innovation. In turn, the absence of financial arrangements that enhance corporate governance may impede the mobilization of savings from disparate agents and also keep capital from flowing to profitable investments [Stiglitz and Weiss (1983)]. Thus, the effectiveness of corporate governance mechanisms directly impacts firm performance with potentially large ramifications on national growth rates. Diffuse shareholders may exert effective corporate governance directly by voting on crucial issues, such as mergers, liquidations, and fundamental changes in business strategies, and indirectly by electing boards of directors to represent the interest of the 3 Indeed, standard agency theory defines the corporate governance problem in terms of how equity and debt holders influence managers to act in the best interests of the providers of capital [e.g., Coase (1937), Jensen and Meckling (1976), Fama and Jensen (1983a, 1983b), Myers and Majluf (1984)].
Ch. 12:
Finance and Growth: Theory and Evidence
873
owners and oversee the myriad of managerial decisions. With low information costs, shareholders can make informed decisions and vote accordingly. In the absence of large market frictions and distorted incentives, boards of directors will represent the interest of all shareholders, oversee managers effectively, and improve the allocation of resources. Starting from at least Berle and Means (1932), however, many researchers have argued that small, diffuse equity may encounter a range of barriers to exerting sound control over corporations.4 An assortment of market frictions, however, may keep diffuse shareholders from effectively exerting corporate governance, which allows managers to pursue projects that benefit themselves rather than the firm and society at large.5 In particular, large information asymmetries typically exist between managers and small shareholders and managers have enormous discretion over the flow of information. Furthermore, small shareholders frequently lack the expertise and incentives to monitor managers because of the large costs and complexity associated with overseeing mangers and exerting corporate control. This may induce a “free-rider” problem because each stockowner’s stake is so small: Each investor relies on others to undertake the costly process of monitoring managers, so there is too little monitoring. The resultant gap in information between corporate insiders and diffuse shareholders implies that the voting rights mechanism will not work effectively. Also, the board of directors may not represent the interests of minority shareholders. Management frequently “captures” the board and manipulates directors into acting in the best interests of the managers, not the shareholders. Finally, in many countries legal codes do not adequately protect the rights of small shareholders and legal systems frequently do not enforce the legal codes that actually are on the books concerning diffuse shareholder rights. Thus, large information and contracting costs may keep diffuse shareholders from effectively exerting corporate governance, with adverse effects on resource allocation and economic growth. One response to the frictions that prevent dispersed shareholders from effectively governing firms is for firms to have a large, concentrated owner, but this ownership structure has its own problems. Large owners have greater incentives to acquire information and monitor managers and greater power to thwart managerial discretion [Grossman and Hart (1980, 1986); Shleifer and Vishny (1996); and Stulz (1988)]. The existence of large shareholders, however, creates a different agency problem: Conflicts arise between the controlling shareholder and other shareholders [Jensen and Meckling (1976)]. The controlling owner may expropriate resources from the firm, or provide jobs, perquisites, and generous business deals to related parties in a manner that hurts the firm and society, but benefits the controlling owner. Indeed, Morck, Wolfenzon and Yeung (2005) show that concentrated ownership appears to have enduring political and
4 In the case of the United States, Roe (1994) argues that the corporate structure of the firms has been heavily
influenced by politics and therefore is not primarily an outcome of market forces. 5 For citations and an insightful discussion, see the review of the corporate governance literature by Shleifer and Vishny (1997).
874
R. Levine
macroeconomic implications. Around the world, controlling owners are frequently powerful families that use pyramidal structures, cross-holdings, and super voting rights to magnify their control over many corporations and banks [La Porta et al. (1999), Morck, Stangeland and Yeung (2000), Claessens et al. (2002), Caprio, Laeven and Levine (2003)]. Morck, Wolfenzon and Yeung (2005) marshal an abundance of evidence in arguing that (i) these controlling families frequently translate their corporate power into political influence and (ii) the elite then use their influence to shape public policies in ways that protect them from competition and subsidize their ventures. Thus, highly concentrated ownership can distort corporate decisions and national policies in ways that curtail innovation, encourage rent-seeking, and stymie economic growth. To the extent that diffuse or concentrated shareholders do not ameliorate the corporate governance problem, theory suggests that other types of financial arrangements may arise to ease market frictions and improve the governance of corporations. In what follows, I discuss how various financial arrangements – liquid equity markets, debt contracts, and banks – may arise to enhance corporate governance and accelerate growth. There are countervailing arguments, however, that each of these financial arrangements actually exerts a deleterious influence on corporate governance. I provide a more complete pro and con assessment of these different mechanisms below when I discuss the bank-based versus market-based debate. Besides the mechanisms discussed thus far, a large and influential literature trumpets the importance of well functioning stock markets in fostering corporate governance [Jensen and Meckling (1976)]. For example, public trading of shares in stock markets that efficiently reflect information about firms allows owners to link managerial compensation to stock prices. Linking stock performance to manager compensation helps align the interests of managers with those of owners [Diamond and Verrecchia (1982) and Jensen and Murphy (1990)]. Similarly, if takeovers are easier in well-developed stock markets and if managers of under-performing firms are fired following a takeover, then better stock markets can promote better corporate control by easing takeovers of poorly managed firms. The threat of a takeover will help align managerial incentives with those of the owners [Scharfstein (1988) and Stein (1988)]. Many, however, argue that well functioning stock markets actually hurt corporate governance. I discuss this below when reviewing the bank-based versus market-based debate. Finally, I am not aware of models that assess the role of markets in boosting steady-state growth through its impact on corporate governance. Some theoretical models indicate that debt contracts may emerge to improve corporate governance, with beneficial ramifications on economic growth. An extensive literature demonstrates how debt contracts may arise to lower the costs of monitoring firm insiders [e.g., Townsend (1979), Gale and Hellwig (1985), Boyd and Smith (1994)]. In terms of growth, Aghion, Dewatripont and Rey (1999) link the use of debt contracts to growth. Using Jensen’s “free cash flow argument”, Aghion, Dewatripont and Rey (1999) show that debt instruments reduce the amount of free cash available to firms. This in turn reduces managerial slack and accelerates the rate at which managers adopt new technologies.
Ch. 12:
Finance and Growth: Theory and Evidence
875
In terms of intermediaries, Diamond (1984) develops a model in which a financial intermediary improves corporate governance. The intermediary mobilizes the savings of many individuals and lends these resources to firms. This “delegated monitor” economizes on aggregate monitoring costs and eliminates the free-rider problem since the intermediary does the monitoring for all the investors. Furthermore, as financial intermediaries and firms develop long-run relationships, this can further lower information acquisition costs. In terms of economic growth, a number of models show that well-functioning financial intermediaries influence growth by boosting corporate governance. Bencivenga and Smith (1993) show that financial intermediaries that improve corporate governance by economizing on monitoring costs will reduce credit rationing and thereby boost productivity, capital accumulation, and growth. Sussman (1993) and Harrison, Sussman and Zeira (1999) develop models where financial intermediaries facilitate the flow of resources from savers to investors in the presence of informational asymmetries with positive growth effects. Focusing on innovative activity, De la Fuente and Marin (1996) develop a model in which intermediaries arise to undertake the particularly costly process of monitoring innovative activities. This improves credit allocation among competing technology producers with positive ramifications on economic growth. From a different perspective, Boyd and Smith (1992) show that differences in the quality of financial intermediation across countries can have huge implications for international capital flows and hence economic growth rates. They show that capital may flow from capital scarce countries to capital abundant countries if the capital abundant countries have financial intermediaries that are sufficiently more effective at exerting corporate control than the capital scarce regions. Thus, even though the physical product of capital is higher in the capital scarce countries, investors recognize that their actual returns depend crucially on the monitoring performed by intermediaries. Thus, poor financial intermediation will lead to sub-optimal allocation of capital. 2.4. Risk amelioration With information and transactions costs, financial contracts, markets and intermediaries may arise to ease the trading, hedging, and pooling of risk with implications for resource allocation and growth. I divide the discussion into three categories: cross-sectional risk diversification, intertemporal risk sharing, and liquidity risk. Traditional finance theory focuses on cross-sectional diversification of risk. Financial systems may mitigate the risks associated with individual projects, firms, industries, regions, countries, etc. Banks, mutual funds, and securities markets all provide vehicles for trading, pooling, and diversifying risk. The financial system’s ability to provide risk diversification services can affect long-run economic growth by altering resource allocation and savings rates. The basic intuition is straightforward. While savers generally do not like risk, high-return projects tend to be riskier than low-return projects. Thus, financial markets that make it easier for people to diversify risk tend to induce a portfolio shift toward projects with higher expected returns [Gurley and Shaw (1955), Patrick (1966),
876
R. Levine
Greenwood and Jovanovic (1990), Saint-Paul (1992), Devereux and Smith (1994) and Obstfeld (1994)].6 Acemoglu and Zilibotti (1997) carefully model the links between cross-sectional risk, diversification, and growth. They assume that (i) high-return, risky projects are frequently indivisible and require a large initial investment, (ii) people dislike risk, (iii) there are lower-return, safe projects, and (iv) capital is scare. In the absence of financial arrangements that allow agents to hold diversified portfolios, agents will avoid the high-return, risky projects because they require agents to invest disproportionately in risky endeavors. They show that financial systems that allow agents to hold a diversified portfolio of risky projects foster a reallocation of savings toward high-return ventures with positive repercussions on growth. In terms of technological change, King and Levine (1993b) show that cross-sectional risk diversification can stimulate innovative activity. Agents are continuously trying to make technological advances to gain a profitable market niche. Engaging in innovation is risky, however. The ability to hold a diversified portfolio of innovative projects reduces risk and promotes investment in growth-enhancing innovative activities (with sufficiently risk averse agents). Thus, financial systems that ease risk diversification can accelerate technological change and economic growth. Besides cross-sectional risk diversification, financial systems may improve intertemporal risk sharing. In examining the connection between cross-sectional risk sharing and growth, theory has tended to focus on the role of markets, rather than intermediaries. However, in examining intertemporal risk sharing, theory has focused on the advantageous role of intermediaries in easing intertemporal risk smoothing [Allen and Gale (1997)]. Risks that cannot be diversified at a particular point in time, such as macroeconomic shocks, can be diversified across generations. Long-lived intermediaries can facilitate intergenerational risk sharing by investing with a long-run perspective and offering returns that are relatively low in boom times and relatively high in slack times. While this type of risk sharing is theoretically possible with markets, intermediaries may increase the feasibility of intertemporal risk sharing by lowering contracting costs. A third type of risk is liquidity risk. Liquidity reflects the cost and speed with which agents can convert financial instruments into purchasing power at agreed prices. Liquidity risk arises due to the uncertainties associated with converting assets into a medium of exchange. Informational asymmetries and transaction costs may inhibit liquidity and intensify liquidity risk. These frictions create incentives for the emergence of financial markets and institutions that augment liquidity. The standard link between liquidity and economic development arises because some high-return projects require a long-run commitment of capital, but savers do not like to relinquish control of their savings for long-periods. Thus, if the financial system does not augment the liquidity of long-term investments, less investment is likely to occur
6 Though not focused on the endogenous emergence of financial markets, Krebs (2003) shows that imperfect sharing of individual human-capital risks can retard long-run economic growth.
Ch. 12:
Finance and Growth: Theory and Evidence
877
in the high-return projects. Indeed, Hicks (1969, pp. 143–145) argues that the products manufactured during the first decades of the Industrial Revolution had been invented much earlier. Rather, the critical innovation that ignited growth in 18th century England was capital market liquidity. With liquid capital markets, savers can hold liquid assets – like equity, bonds, or demand deposits – that they can quickly and easily sell if they seek access to their savings. Simultaneously, capital markets transform these liquid financial instruments into long-term capital investments. Thus, the industrial revolution required a financial revolution so that large commitments of capital could be made for long periods [Bencivenga, Smith and Starr (1995)]. In Diamond and Dybvig’s (1983) seminal model of liquidity, a fraction of savers receive shocks after choosing between two investments: an illiquid, high-return project and a liquid, low-return project. Those receiving shocks want access to their savings before the illiquid project produces. This risk creates incentives for investing in the liquid, low-return projects. The model assumes that it is prohibitively costly to verify whether another individual has received a shock or not. This information cost assumption rules out state-contingent insurance contracts and creates an incentive for financial markets – markets where individuals issue and trade securities – to emerge. Levine (1991) takes the Diamond and Dybvig (1983) set-up, models the endogenous formation of equity markets, and links this to a growth model. Specifically, savers receiving shocks can sell their equity claims to the future profits of the illiquid production technology to others. Market participants do not verify whether other agents received shocks or not. Participants simply trade in impersonal stock exchanges. Thus, with liquid stock markets, equity holders can readily sell their shares, while firms have permanent access to the capital invested by the initial shareholders. By facilitating trade, stock markets reduce liquidity risk. Frictionless stock markets, however, do not eliminate liquidity risk. That is, stock markets do not replicate the equilibrium that exists when insurance contracts can be written contingent on observing whether an agent receives a shock or not. Nevertheless, as stock market transaction costs fall, more investment occurs in the illiquid, high-return project. If illiquid projects enjoy sufficiently large externalities, then greater stock market liquidity induces faster steady-state growth. Thus far, information costs – the costs of verifying whether savers have received a shock – have motivated the existence of stock markets, but trading costs can also hasten the emergence and highlight the importance of liquid stock markets. In Bencivenga, Smith and Starr (1995), high-return, long-gestation production technologies require that ownership be transferred throughout the life of the production process in secondary securities markets. If exchanging ownership claims is costly, then longer-run production technologies will be less attractive. Thus, liquidity – as measured by secondary market trading costs – affects production decisions. Greater liquidity will induce a shift to longer-gestation, higher-return technologies. Besides stock markets, financial intermediaries may also enhance liquidity, reduce liquidity risk and influence economic growth. As discussed above, Diamond and Dybvig’s (1983) model assumes it is prohibitively costly to observe shocks to individuals, so
878
R. Levine
it is impossible to write incentive compatible state-contingent insurance contracts. Under these conditions, banks can offer liquid deposits to savers and undertake a mixture of liquid, low-return investments to satisfy demands on deposits and illiquid, high-return investments. By providing demand deposits and choosing an appropriate mixture of liquid and illiquid investments, banks provide complete insurance to savers against liquidity risk while simultaneously facilitating long-run investments in high return projects. Banks replicate the equilibrium allocation of capital that exists with observable shocks. As noted by Jacklin (1987), however, the banking equilibrium is not incentive compatible if agents can trade in liquid equity markets. If equity markets exist, all agents will use equities; none will use banks. Thus, in this context, banks will only emerge to provide liquidity if there are sufficiently large impediments to trading in securities markets [Diamond (1991)].7 Turning back to growth, Bencivenga and Smith (1991) examine a growth model in which pre-existing impediments to the emergence of liquid equity markets highlight the liquidity-enhancing role of banks. They show that, by eliminating liquidity risk, banks can increase investment in the high-return, illiquid asset and therefore accelerate growth. Financial systems can also promote the accumulation of human capital [Jacoby (1994)]. In particular, financial arrangements may facilitate borrowing for the accumulation of skills. If human capital accumulation is not subject to diminishing returns on a social level, financial arrangements that ease human capital creation help accelerate economic growth [De Gregorio (1996), Galor and Zeira (1993)]. Another form of liquidity involves firm access to credit during the production process, which may reduce premature liquidity of projects and thereby foster investment in longer gestation, higher-return projects. Holmstrom and Tirole (1998) note that firm production processes are long-term, uncertain, and subject to shocks. Thus, some firms may receive shocks after receiving outside financing and need additional injections of capital to complete the project. In the presence of informational asymmetries, intermediaries can sell an option to a line of credit during the initial financing of the firm that entitles the firm to access additional credit at an intermediate stage in certain states of nature.8 This improves the efficiency of the capital allocation process, but the model does
7 Note that Calomiris and Kahn (1991), Flannery (1994), and Diamond and Rajan (2001) develop models in which the fragile structure of banks, i.e., liquid deposits and illiquid assets, serves as an effective commitment device that keeps banks from assuming excessive risks or from shirking on collecting payment from firms. Put succinctly, the sequential service constraint on bank deposits creates a collective action problem among depositors that induces depositors to run if they acquire information that the bank is not monitoring firms and managing risk appropriately. 8 Kashyap, Stein and Rajan (2002) develop a model to explain why two traditional commercial banking activities, deposit-taking and loan-making, are done jointly within the same intermediary. While Diamond (1984) motivates the existence of loan-making intermediaries, it is not clear why the intermediary should be funded with demand deposits. Similarly, while Gorton and Pennacchi (1990) motivate the existence of deposit-taking intermediaries, it is not clear why the intermediary should make loans. Since banks often lend via committed lines of credit that can be used at the discretion of firms, Kashyap, Stein and Rajan (2002) argue that a single bank is frequently the lowest cost provider of liquidity on both the liability and asset sides
Ch. 12:
Finance and Growth: Theory and Evidence
879
not formally link the provision of liquidity with economic growth. Aghion et al. (2004), instead, focus on how the ability of firms to access credit during the production process influences innovation and long-run growth when firms face macroeconomic shocks (e.g., recessions). They develop a model where firms can either invest in short-term, low-return investments or in more risky, growth-enhancing research and development (R&D). They also assume that there are adjustment costs to R&D. In this context, underdeveloped financial systems that are less able to provide firms with funds to ease these adjustment costs will hinder innovation. Moreover, macroeconomic volatility exerts a particularly negative impact on innovation and growth in under-developed financial systems because firms’ willingness to undertake R&D depends on their ability to borrow in the future to meet adjustment costs, which is influenced negatively by the likelihood of experiencing a recession and positively by the level of financial development. Aghion et al. (2004) also provide empirical evidence consistent with the prediction that financial development reduces the adverse growth effects of macroeconomic volatility. 2.5. Pooling of savings Mobilization – pooling – is the costly process of agglomerating capital from disparate savers for investment. Mobilizing savings involves (a) overcoming the transaction costs associated with collecting savings from different individuals and (b) overcoming the informational asymmetries associated with making savers feel comfortable in relinquishing control of their savings. Indeed, much of Carosso’s (1970) history of Investment Banking in America is a description of the diverse costs associated with raising capital in the United States during the 19th and 20th centuries. In light of the transaction and information costs associated with mobilizing savings from many agents, numerous financial arrangements may arise to mitigate these frictions and facilitate pooling. Specifically, mobilization may involve multiple bilateral contracts between productive units raising capital and agents with surplus resources. The joint stock company in which many individuals invest in a new legal entity, the firm, represents a prime example of multiple bilateral mobilizations. To economize on the costs associated with multiple bilateral contracts, pooling may also occur through intermediaries, where thousands of investors entrust their wealth to intermediaries that invest in hundreds of firms [Sirri and Tufano (1995, p. 83)]. For this to occur, “mobilizers” have to convince savers of the soundness of the investments [Boyd and Smith (1992)]. Toward this end, intermediaries worry about establishing stellar reputations, so that savers feel comfortable about entrusting their savings to the intermediary [DeLong (1991) and Lamoreaux (1994)]. Financial systems that are more effective at pooling the savings of individuals can profoundly affect economic development by increasing savings, exploiting economies
of the balance sheet. This view of intermediation has not, to my knowledge, been incorporated into a model of economic growth.
880
R. Levine
of scale, and overcoming investment indivisibilities. Besides the direct effect of better savings mobilization on capital accumulation, better savings mobilization can improve resource allocation and boost technological innovation. Without access to multiple investors, many production processes would be constrained to economically inefficient scales [Sirri and Tufano (1995)]. Furthermore, many endeavors require an enormous injection of capital that is beyond the means or inclination of any single investor. Bagehot (1873, pp. 3–4) argued that a major difference between England and poorer countries was that in England the financial system could mobilize resources for “immense works”. Thus, good projects would not fail for lack of capital. Bagehot was very explicit in noting that it was not the national savings rate per se, it was the ability to pool society’s resources and allocate those savings toward the most productive ends. Furthermore, mobilization frequently involves the creation of small denomination instruments. These instruments provide opportunities for households to hold diversified portfolios [Sirri and Tufano (1995)]. Acemoglu and Zilibotti (1997) show that with large, indivisible projects, financial arrangements that mobilize savings from many diverse individuals and invest in a diversified portfolio of risky projects facilitate a reallocation of investment toward higher return activities with positive ramifications on economic growth. 2.6. Easing exchange Financial arrangements that lower transaction costs can promote specialization, technological innovation and growth. The links between facilitating transactions, specialization, innovation, and economic growth were core elements of Adam Smith’s (1776) Wealth of Nations. He argued that division of labor – specialization – is the principal factor underlying productivity improvements. With greater specialization, workers are more likely to invent better machines or production processes [Smith (1776, p. 3)]. Men are much more likely to discover easier and readier methods of attaining any object, when the whole attention of their minds is directed towards that single object, than when it is dissipated among a great variety of things. Smith (1776) focused on the role of money in lowering transaction costs, permitting greater specialization, and fostering technological innovation.9 Information costs, however, may also motivate the emergence of money. Since it is costly to evaluate the attributes of goods, barter exchange is very costly. Thus, an easily recognizable medium of exchange may arise to facilitate exchange [King and Plosser (1986) and Williamson and Wright (1994)]. The drop in transaction and information costs is not necessarily a one-time fall when economies move to money, however. Transaction and information costs may continue to fall through financial innovation.
9 Wright (2002, pp. 212–216) documents that Adam Smith, in the second book of the Wealth of Nations, indicated that well-functioning banks, besides well-functioning corporate governance mechanisms, were crucial for economic development.
Ch. 12:
Finance and Growth: Theory and Evidence
881
Greenwood and Smith (1996) have modeled the connections between exchange, specialization, and innovation. More specialization requires more transactions. Since each transaction is costly, financial arrangements that lower transaction costs will facilitate greater specialization. In this way, markets that promote exchange encourage productivity gains. There may also be feedback from these productivity gains to financial market development. If there are fixed costs associated with establishing markets, then higher income per capita implies that these fixed costs are less burdensome as a share of per capita income. Thus, economic development can spur the development of financial markets. In the Greenwood and Smith (1996) model, however, the reduction in transaction costs does not stimulate the invention of new and better production technologies. Instead, lower transaction costs expands the set of “on the shelf” production processes that are economically attractive. Also, the model defines better “market” as a system for supporting more specialized production processes. This does not explain the emergence of financial instruments or institutions that lower transactions costs and thereby produce an environment that naturally promotes specialized production technologies. This is important because we want to understand the two links of the chain: what features of the economic environment create incentives for the emergence of financial arrangements, and how do these emerging financial arrangements influence economic activity. 2.7. The theoretical case for a bank-based system Besides debates concerning the role of financial development in economic growth, financial economists have debated the comparative importance of bank-based and market-based financial systems for over a century [Goldsmith (1969), Boot and Thakor (1997), Allen and Gale (2000), Demirgüç-Kunt and Levine (2001c)]. As discussed, financial intermediaries can improve the (i) acquisition of information on firms, (ii) intensity with which creditors exert corporate control, (iii) provision of risk-reducing arrangements, (iv) pooling of capital, and (v) ease of making transactions. These are arguments in favor of well-developed banks. They are not reasons for favoring a bankbased financial system. Rather than simply noting the growth-enhancing role of banks, the case for a bankbased system derives from a critique of the role of markets in providing financial functions. In terms of acquiring information about firms, Stiglitz (1985) emphasizes the freerider problem inherent in atomistic markets. Since well-developed markets quickly reveal information to investors at large, this dissuades individual investors from devoting resources toward researching firms. Thus, greater market development, in lieu of bank development, may actually impede incentives for identifying innovative projects that foster growth.10 Banks can mitigate the potential disincentives from efficient markets 10 Using examples from the U.S. in the 18th Century, Wright (2002, pp. 30–32) shows how securities market
participants tended to free-ride off of the information collected by banks in making credit decisions.
882
R. Levine
by privatizing the information they acquire and by forming long-run relationships with firms [Gerschenkron (1962), Boot, Greenbaum and Thakor (1993)]. Banks can make investments without revealing their decisions immediately in public markets and this creates incentives for them to research firms, managers, and market conditions with positive ramifications on resource allocation and growth. Furthermore, Rajan and Zingales (1999) emphasize that powerful banks with close ties to firms may be more effective at exerting pressure on firms to re-pay their debts than atomistic markets. On corporate governance, a large literature stresses that markets do not effectively monitor managers [Shleifer and Vishny (1997)]. First, takeovers may not be an effective corporate control device because insiders have better information than outsiders. This informational asymmetry mitigates the takeover threat as a corporate governance mechanism since ill-informed outsiders will outbid relatively well-informed insiders for control of firms only when they pay too much [Stiglitz (1985)]. Second, some argue that the takeover threat as a corporate control device also suffers from the free-rider problem. If an outsider expends lots of resources obtaining information, other market participants will observe the results of this research when the outsider bids for shares of the firm. This will induce others to bid for shares, so that the price rises. Thus, the original outsider who expended resources obtaining information must pay a higher price for the firm than it would have paid if “free-riding” firms could not bid for shares in a liquid equity market. The rapid public dissemination of costly information reduces incentives for obtaining information, making effective takeover bids, and wielding corporate control [Grossman and Hart (1980)]. Third, existing managers often take actions – poison pills – that deter takeovers and thereby weaken the market as an effective disciplining device [DeAngelo and Rice (1983)]. There is some evidence that, in the United States, the legal system hinders takeovers and grants considerable power to management. Fourth, although in theory shareholder control management through boards of directors, an incestuous relationship may blossom between boards of directors and management [Jensen (1993)]. Members of a board enjoy their lucrative fees and owe those fees to nomination by management. Thus, boards are more likely to approve golden parachutes to managers and poison pills that reduce the attractiveness of takeover. This incestuous link may further reduce the effectiveness of the market as a vehicle for exerting corporate control [Allen and Gale (2000)]. Chakraborty and Ray (2004) examine bank-based and market-based financial systems in an endogenous growth model, concluding that banks can partially resolve the tendency for insiders to exploit the private benefits of control. The liquidity of stock markets can also adversely influence resource allocation. Liquid equity markets may facilitate takeovers that while profiting the raiders may actually be socially harmful [Shleifer and Summers (1988)]. Moreover, liquidity may encourage a myopic investor climate. In liquid markets, investor can inexpensively sell their shares, so that they have fewer incentives to undertake careful – and expensive – corporate governance [Bhide (1993)]. Thus, greater stock market development may hinder corporate governance and induce an inefficient allocation of resources according to the bank-based view. As noted above, Allen and Gale (1997, 2000) argue that bank-based
Ch. 12:
Finance and Growth: Theory and Evidence
883
systems offer better intertemporal risk sharing services than markets with beneficial effects on resource allocation. In response to the problems associated with relying on diffuse shareholders, large, concentrated ownership may arise to prevent managers from deviating too far from the interests of owners, but as stressed above, this brings its own complications. Large investors have the incentives and ability to acquire information, monitor managers and exert corporate control. Concentrated ownership, however, raises other problems. Besides the fact that concentrated ownership implies that wealthy investors are not diversified [Acemoglu and Zilibotti (1997)], concentrated owners may benefit themselves at the expense of minority shareholders, debt holders, and other stakeholders in the firm, with adverse effects on corporate finance and resource allocation. Large investors may pay themselves special dividends, exploit business relationships with other firms they own that profit themselves at the expense of the corporation, and in general maximize the private benefits of control at the expense of minority shareholders [Zingales (1994)]. Furthermore, large equity owners may seek to shift the assets of the firm to higher-risk activities since shareholders benefit on the upside, while debt holders share the costs of failure. Finally, as stressed by Morck, Wolfenzon and Yeung (2005), concentrated control of corporate assets produces market power that may corrupt the political system and distort public policies. Thus, from this perspective, concentrated ownership is unlikely to resolve fully the shortcomings associated with market-based systems. In sum, proponents of bank-based systems argue that there are fundamental reasons for believing that market-based systems will not do a good job of acquiring information about firms and overseeing managers. This will hurt resource allocation and economic performance. Banks do not suffer from the same fundamental shortcomings as markets. Thus, they will do a correspondingly better job at researching firms, overseeing managers, and financing industrial expansion. 2.8. The theoretical case for a market-based system The case for a market-based system is essentially a counterattack that focuses on the problems created by powerful banks. Bank-based systems may involve intermediaries with a huge influence over firms and this influence may manifest itself in negative ways. For instance, once banks acquire substantial, inside information about firms, banks can extract rents from firms; firms must pay for their greater access to capital. In terms of new investments or debt renegotiations, banks with power can extract more of the expected future profits from the firm (than in a market-based system) [Hellwig (1991)]. This ability to extract part of the expected payoff to potentially profitable investments may reduce the effort extended by firms to undertake innovative, profitable ventures [Rajan (1992)]. Furthermore, Boot and Thakor (2000) model the potential tensions between bank-based systems characterized by close ties between banks and firms and the development of well-functioning securities markets.
884
R. Levine
Banks – as debt issuers – also have an inherent bias toward prudence, so that bankbased systems may stymie corporate innovation and growth [Morck and Nakamura (1999)]. Weinstein and Yafeh (1998) find evidence of this in Japan. While firms with close to ties to a “main bank” have greater access to capital and are less cash constrained than firms without a main bank, the main bank firms tend to (i) employ conservative, slow growth strategies and do not grow faster than firms without a “main bank”, (ii) use more capital intensive processes than non-main bank firms holding other features constant, and (iii) produce lower profits, which is consistent with the powerful banks extracting rents from the relationship. Allen and Gale (2000) further note that although banks may be effective at eliminating duplication of information gathering and processing, which is likely to be helpful when people agree about what information needs to be gathered and how it should be processed, banks may be ineffective in non-standard environments. Thus, banks may not be effective gatherers and processors of information in new, uncertain situations involving innovative products and processes [Allen and Gale (1999)]. Similarly, but in a model of loan renegotiations, Dewatripont and Maskin (1995) demonstrate that in a bank-based system characterized by long-run links between banks and firms, banks will have a difficult time credibly committing to not renegotiate contracts. In contrast, more fragmented banking systems can more easily commit to imposing tighter budget constraints. The credible imposition of tight budget constraints may be necessary for the funding of newer, higher-risk firms. Thus, concentrated banks may be more conducive to the funding of mature, less risky firms, while more market-based systems, according to these theories, more easily support the growth of newer, riskier industries. Another line of attack on the efficacy of bank-based systems involves their role in exerting corporate control over firms and the corporate governance of banks themselves. Bankers act in their own best interests, not necessarily in the best interests of all creditors or society at large. Thus, bankers may collude with firms against other creditors. For instance, influential banks may prevent outsiders from removing inefficient managers if these managers are particularly generous to the bankers [Black and Moersch (1998)].11 For the case of Germany, Wenger and Kaserer (1998) show that bank managers are enormously powerful. They not only have the corporate control power over firms that derives from being large creditors to those firms, banks also vote the shares of a larger number of small stockholders. For instance, in 1992, bank managers exercised on average 61 percent of the voting rights of the 24 largest companies and in 11 companies this share was higher than 75%. This control of corporations by bank management extends to the banks themselves! In the shareholder meetings of the three largest German banks, the percentage of proxy votes was higher than 80 percent, much of this voted by the banks themselves. For example, Deutsche Bank held voting rights for 47 percent of its own
11 Bank-based system may also impede the flow of information about firms [Morck, Stangeland and Yeung
(2000)] and the responsiveness of the economy to market signals [Hoshim, Kashyap and Sharfstein (1991), Peek and Rosengren (1998)].
Ch. 12:
Finance and Growth: Theory and Evidence
885
shares, while Dresdner votes 59 percent of its own shares [Charkham (1994)]. Thus, the bank management has rested control of the banks from the owners of the banks and also exerts a huge influence on the country’s major corporations. Wenger and Kaserer (1998) also provide examples in which banks misrepresent the accounts of firms to the public and systematically fail to discipline management. Also, Rajan and Zingales (2003) argue that in response to adverse shocks that affect the economy unevenly, market-based systems will more effectively identify, isolate, and bankrupt truly distressed firms and prevent them from hurting the overall economy than a bank-based system. In a bankbased – relationship-based – system, bank managers may be more reluctant to bankrupt firms with whom they have had long-term, and perhaps multidimensional, ties. While this may smooth temporary aggregate shocks, it may also impede the efficient adjustment to structural changes. Thus, to the extent that banks actually weaken the corporate governance of firms, bank-based systems represent sub-optimal mechanisms for overseeing firms and improving resource allocation. Furthermore, relying on a bank-based financial system may be problematic because of the difficulties in governing banks themselves [Caprio and Levine (2002)]. While subject to debate, many argue that information asymmetries between bank insiders and outsiders are larger than with nonfinancial corporations [Furfine (2001), Morgan (2002)]. Under these conditions, it will be very difficult for diffuse equity and debt holder to monitor and control bank insiders [Laeven and Levine (2005)]. The governance problem facing depositors is of course exacerbated in the presence of deposit insurance. Furthermore, greater opacity implies even greater complexities in writing incentive contracts to align managerial incentives with bank equity holders and creditors. Perhaps because of the particularly severe informational impediments to governing banks, banks are even more likely than nonfinancial corporations to have a large, controlling owner [Caprio, Laeven and Levine (2003)]. This concentration of ownership in conjunction with greater opaqueness may make it easier for bank insiders to exploit both other investors in the bank and the government if it is providing deposit insurance. The history of Mexico, for example, is replete with incidents of powerful families using their control over banks to exploit other creditors and taxpayers [Haber (2004, 2005), Maurer and Haber (2004)]. For instance, La Porta, Lopez-de-Silanes and Zamarripa (2003) find high rates of connected lending in Mexico. They find that 20% of total loans go to related parties. These loans benefited from interest rates that were about 415–420 basis points below those to unrelated parties. Related borrowers also benefited from longer maturities, were significantly less likely to have to post collateral, were 33% less likely to pay back, and the recovery rates on these loans were massively less (78 percent lower) than on loans to unrelated parties. Similarly, Laeven (2001) presents evidence that insiders in Russian banks diverted the flow of loans to themselves and then defaulted 71% of the time. Finally, proponents of market-based financial systems claim that markets provide a richer set of risk management tools that permit greater customization of risk ameliorating instruments. While bank-based systems may provide inexpensive, basic risk management services for standardized situations, market-based systems provide greater
886
R. Levine
flexibility to tailor make products. Thus, as economies mature and need a richer set of risk management tools and vehicles for raising capital, they may concomitantly benefit from a legal and regulatory environment that supports the evolution of market-based activities, or overall growth may be retarded. 2.9. Countervailing views to bank-based vs. market-based debate Some reject the importance of the bank-based versus market-based debate and instead argue that the issue is overall financial development, not the particular institutional arrangements that provide financial services to the economy. As noted above, information, transaction, and enforcement costs create incentives for the emergence of financial markets and intermediaries. In turn, these components of the financial system provide financial functions: they evaluate project, exert corporate control, facilitate risk management, ease the mobilization of savings, and facilitate exchange. Thus, this “financial functions view” rejects the primacy of distinguishing financial systems as bank-based or market-based [Merton (1992, 1995), Merton and Bodie (1995, 2004), Levine (1997)]. According to this view, the crucial issue for growth is whether the economy has access to a well-functioning financial system; the exact composition of the financial system is of secondary importance. Another criticism for emphasizing market-based versus bank-based differences is that markets and banks may provide complementary growth-enhancing financial services to the economy [Boyd and Smith (1998), Levine and Zervos (1998a), Huybens and Smith (1999)]. For instance, stock markets may positively affect economic development even though not much capital is raised through them. Specifically, stock markets may play a prominent role in facilitating custom-made risk management services and boosting liquidity. In addition, stock markets may complement banks. For instance, by spurring competition for corporate control and by offering alternative means of financing investment, securities markets may reduce the potentially harmful effects of excessive bank power. The theoretical literature is making progress in modeling the co-evolution of banks and markets [Boyd and Smith (1996), Allen and Gale (2000)]. Furthermore, microeconomic evidence also emphasizes potential complementarities between intermediaries and markets. Using firm-level data, Demirgüç-Kunt and Maksimovic (1996) show that increases in stock market development actually tend to increase the use of bank finance in developing countries. Moreover, Sylla (1998) describes the interdependence of banks and securities markets in providing financial services to the U.S. economy in the late 18th and early 19th centuries. Thus, these two components of the financial system may act as complements during the development process. In many circumstances, we may not want to view bank-based and market-based systems as representing a tradeoff. Rather, there may be policy and analytical advantages to focusing on the legal, regulatory, and policy that allow both banks and markets to flourish without tipping the playing field in favor of either banks or markets. One additional argument for not focusing on distinguishing financial systems by whether they are bank-based or market-based is the view that legal system differences
Ch. 12:
Finance and Growth: Theory and Evidence
887
are the fundamental source of international differences in financial development [La Porta et al. (2000)]. The law and finance view holds that finance is a set of contracts. These contracts are defined and made more or less effective by legal rights and enforcement mechanisms. From this perspective, a well-functioning legal system facilitates the operation of both markets and intermediaries. It is the overall level and quality of the financial functions that are provided to the economy that influences resource allocation and economic growth. The law and finance view holds that distinguishing countries by the efficiency of national legal systems in supporting financial transactions is more useful than distinguishing countries by whether they have bank-based or market-based financial systems. While focusing on the law is not inconsistent with banks or markets playing a particularly important role, La Porta et al. (2000) clearly argue that legal institutions are a more useful way to distinguish financial systems than concentrating on whether countries are bank-based or market-based. 2.10. Finance, income distribution, and poverty Thus far, I have focused on models of aggregate growth. I have not discussed the potential impact of finance on income distribution in general or poverty in particular. Although the focus of this article is on aggregate growth, the relationship between finance and income distribution is independently relevant for understanding the process of economic development and is indirectly related to growth because income distribution can influence savings decisions, the allocation of resources, incentives to innovate, and public policies. Thus, this subsection very briefly reviews a few recent theoretical inquiries into the relationship between the operation of the financial sector and income distribution. Theory provides conflicting predictions concerning the relationship between financial development and both income distribution and poverty alleviation. Some theories claim that financial intermediary development will have a disproportionately beneficial impact on the poor. Banerjee and Newman (1993), Galor and Zeira (1993) and Aghion and Bolton (1997) show that informational asymmetries produce credit constraints that are particularly binding on the poor because the poor do not have the resources to fund their own projects, nor the collateral (nor the political connections) to access bank credit. These credit constraints, therefore, restrict the poor from exploiting investment opportunities. While these credit constraints may slow aggregate growth by keeping capital from flowing to its highest value use, a poorly functioning financial system will also produce higher income inequality by disproportionately keeping capital from flowing to “wealth-deficient” entrepreneurs. By ameliorating information and transactions costs and therefore by allowing more entrepreneurs to obtain external finance, financial development improves the allocation of capital, exerting a particularly large impact on the poor. On a more general level, some political economy theories suggest that better functioning financial systems make financial services available to a larger proportion of the population, rather than restricting capital to entrenched incumbents [Haber, Maurer and Razo (2003), Rajan and Zingales (2003), Morck, Wolfenzon and Yeung (2005)]. Thus,
888
R. Levine
by ameliorating credit constraints, financial development may foster entrepreneurship, new firm formation, and economic growth. On the other hand, some argue that it is primarily the rich and politically connected who benefit from improvements in the financial system. Especially at early stages of economic development, access to financial services, especially credit, is limited to the wealthy and connected [Lamoreaux (1994), Haber (1991, 2004, 2005)]. Under these conditions, greater financial development may only succeed in channeling more capital to a select few. Thus, it is an open question whether financial development will narrow or widen income disparities even if it boosts aggregate growth. Other models posit a non-linear relationship between finance and income distribution. Greenwood and Jovanovic (1990) show how the interaction of financial and economic development can give rise to an inverted U-shaped curve of income inequality and financial intermediary development. At early stages of financial development, only a few relatively wealthy individuals have access to financial markets and hence higher-return projects. With aggregate economic growth, more people can afford to joint the formal financial system, with positive ramifications on economic growth. With sufficient economic success, everyone participates in the financial system, enjoying the full range of benefits. The distributional effect of financial deepening is thus adverse for the poor at early stages, but positive after a turning point.
3. Evidence on finance and growth A substantial body of empirical work on finance and growth assesses the impact of the operation of the financial system on economic growth, whether the impact is economically large, and whether certain components of the financial system, e.g., banks and stock markets, play a particularly important role in fostering growth at certain stages of economic development. This section is organized around econometric approaches to examining the relationship between finance and growth. Thus, the first subsection discusses cross-country studies of growth and finance. The second subsection presents evidence from panel studies, pure time-series investigations, and country case-studies. The third subsection examines industry and firm level analyses that provide direct empirical evidence on the mechanisms linking finance and growth. Then, I summarize existing work on the relationship between financial structure – the degree to which an economy is bank-based or market-based – and economic growth. Finally, I mention recent research on whether financial development influences income distribution and poverty. The organization of the empirical evidence advertises an important weakness in the finance and growth literature: there is frequently an insufficiently precise link between theory and measurement. Theory focuses on particular functions provided by the financial sector – producing information, exerting corporate governance, facilitating risk management, pooling savings, and easing exchange – and how these influence resource
Ch. 12:
Finance and Growth: Theory and Evidence
889
allocation decisions and economic growth. Thus, I would prefer to organize the empirical section around studies that precisely measure each of the functions stressed by theory. Similarly, while empirical studies focus on measures of the size of banks or stock markets, Petersen and Rajan (1997), Demirgüç-Kunt and Maksimovic (2001), and Fisman and (2003a, 2003b) show that firms frequently act as financial intermediaries in providing trade credit to related firms. This source of financial intermediation may be very important, especially in countries with regulatory restrictions on financial intermediaries and in countries with undeveloped legal systems that do not effectively support formal financial development. This further advertises the sub-optimal connection between theory and measurement in the finance and growth literature. While fully recognizing this problem, many of the biggest advances in empirical studies of finance and growth have been methodological. Thus, I organize the discussion around econometric approaches. While serious improvements have been made in measuring financial development, which I discuss below, future research that more concretely links the concepts from theory with the data will substantively improve our understanding of the finance and growth link. 3.1. Cross-country studies of finance and growth 3.1.1. Goldsmith, the question, and the problems Goldsmith (1969) motivated his path breaking study of finance and growth as follows. One of the most important problems in the field of finance, if not the single most important one, . . . is the effect that financial structure and development have on economic growth. (p. 390) Thus, he sought to assess whether finance exerts a causal influence on growth and whether the mixture of markets and intermediaries operating in an economy influences economic growth. Toward this end, Goldsmith (1969) carefully compiled data on 35 countries over the period 1860 to 1963 on the value of financial intermediary assets as a share of economic output. He assumed, albeit with ample qualifications, that the size of the financial intermediary sector is positively correlated with the quality of financial functions provided by the financial sector. Goldsmith (1969) met with varying degrees of success in providing confident answers to these questions. After showing that financial intermediary size relative to the size of the economy rises as countries develop, Goldsmith graphically documented a positive correlation between financial development and the level of economic activity. Goldsmith just as clearly asserted his unwillingness to draw causal interpretations from his graphical presentations. Thus, Goldsmith ultimately did not take a stand on whether financial development causes growth. In terms of the relationship between economic growth and the structure of the financial system, Goldsmith was unable to provide much cross-country evidence because of the absence of data on securities market development for a broad range of countries.
890
R. Levine
Goldsmith’s (1969) work raises several problems, all of which Goldsmith presciently stresses, that subsequent work has tried to resolve. (1) The investigation involves only 35 countries. (2) It does not systematically control for other factors influencing economic growth. (3) It does not examine whether financial development is associated with productivity growth and capital accumulation, which theory stresses. (4) The indicator of financial development, which measures the size of the financial intermediary sector, may not accurately gauge the functioning of the financial system. (5) The close association between financial system size and growth does not identify the direction of causality. (6) The study did not shed light on whether financial markets, non-bank financial intermediaries, or the mixture of markets and intermediaries matter for economic growth. 3.1.2. More countries, more controls, and predictability In the early 1990s, King and Levine (1993a, henceforth KL) built on Goldsmith’s work. They study 77 countries over the period 1960–1989, systematically control for other factors affecting long-run growth, examine the capital accumulation and productivity growth channels, construct additional measures of the level of financial development, and analyze whether the level of financial development predicts long-run economic growth, capital accumulation, and productivity growth. In terms of measures of financial development, KL first examine DEPTH, which is simply a measure of the size of financial intermediaries. It equals liquid liabilities of the financial system (currency plus demand and interest-bearing liabilities of banks and nonbank financial intermediaries) divided by GDP. They also construct the variable BANK that measures the relative degree to which the central bank and commercial banks allocate credit. BANK equals the ratio of bank credit divided by bank credit plus central bank domestic assets. The intuition underlying this measure is that banks are more likely to provide the five financial functions than central banks. There are two notable weaknesses with this measure, however. Banks are not the only financial intermediaries providing valuable financial functions and banks may simply lend to the government or public enterprises. KL also examine PRIVY, which equals credit to private enterprises divided by GDP. The assumption underlying this measure is that financial systems that allocate more credit to private firms are more engaged in researching firms, exerting corporate control, providing risk management services, mobilizing savings, and facilitating transactions than financial systems that simply funnel credit to the government or state owned enterprises. While BANK and PRIVY seek to improve upon DEPTH by capturing who is doing the allocating and to whom society’s savings are flowing, these measures still do not directly proxy for the five financial functions stressed in theoretical models of finance and growth. KL find very consistent results across the different financial development indicators.
Ch. 12:
Finance and Growth: Theory and Evidence
891
KL then assess the strength of the empirical relationship between each of these indicators of the level of financial development averaged over the 1960–1989 period and three growth indicators also averaged over the 1960–1989 period. The three growth indicators are as follows: (1) the average rate of real per capita GDP growth, (2) the average rate of growth in the capital stock per person, and (3) total productivity growth, which is a “Solow residual” defined as real per capita GDP growth minus (0.3) times the growth rate of the capital stock per person. In other words, if F (i) represents the value of the ith indicator of financial development averaged over the period 1960–1989, G(j ) represents the value of the j th growth indicator (per capita GDP growth, per capita capital stock growth, or productivity growth) averaged over the period 1960–1989, and X represents a matrix of conditioning information to control for other factors associated with economic growth (e.g., income per capita, education, political stability, indicators of exchange rate, trade, fiscal, and monetary policy), then they estimated the following regressions on a cross-section of 77 countries: G(j ) = α + βF (i) + γ X + ε. Table 1 is adapted from KL and indicates that there is a strong positive relationship between each of the financial development indicators, F (i), and the three growth indicators G(i), long-run real per capita growth rates, capital accumulation and productivity growth. The sizes of the coefficients are economically large. Ignoring causality, the coefficient on DEPTH implies that a country that increased DEPTH from the mean of the slowest growing quartile of countries (0.2) to the mean of the fastest growing quartile of countries (0.6) would have increased its per capita growth rate by almost 1 percent per year. This is large. The difference between the slowest growing 25 percent of countries and the fastest growing quartile of countries is about five percent per annum over this 30-year period. Thus, the rise in DEPTH alone eliminates 20 percent of this growth difference. King and Levine (1993b, 1993c) confirm these findings using alternative econometric methods and robustness checks. To examine whether finance simply follows growth, KL study whether the value of financial depth in 1960 predicts the rate of economic growth, capital accumulation, and productivity growth over the next 30 years. Table 2 summarizes these results. The dependent variable is, respectively, real per capital GDP growth, real per capita capital stock growth, and productivity growth averaged over the period 1960–1989. The financial indicator in each of these regressions is the value of DEPTH in 1960. The regressions indicate that financial depth in 1960 is a good predictor of subsequent rates of economic growth, physical capital accumulation, and economic efficiency improvements over the next 30 years even after controlling for income, education, and measures of monetary, trade, and fiscal policy. The relationship between the initial level of financial development and growth is economically large. For example, the estimated coefficients suggest that if in 1960 Bolivia had increased its financial depth from 10 percent of GDP to the mean value for developing countries in 1960 (23 percent), then Bolivia would have grown about 0.4 percent faster per annum, so that by 1990 real per capita GDP would have been about 13 percent larger than it was. These examples
892
R. Levine Table 1 Growth and financial intermediary development, 1960–1989
Dependent variable
DEPTH
BANK
PRIVY
Real per capita GDP growth
2.4∗∗ (0.007) 0.50 2.2∗∗ (0.006) 0.65 1.8∗∗ (0.026) 0.42
3.2∗∗ (0.005) 0.50 2.2∗∗ (0.008) 0.62 2.6∗∗ (0.010) 0.43
3.2∗∗ (0.002) 0.52 2.5∗∗ (0.007) 0.64 2.5∗∗ (0.006) 0.44
R2 Real per capita capital growth R2 Productivity growth R2
Source: King and Levine (1993b, Table VII). ∗ Significant at the 0.10 level. ∗∗ Significant at the 0.05 level. (p-values in parentheses.) Observations: 77. Variable definitions: DEPTH = Liquid liabilities/GDP, BANK = Deposit bank domestic credit/[Deposit bank domestic credit + Central bank domestic credit], PRIVY = Gross claims on the private sector/GDP, Productivity growth = Real per capita GDP growth − 0.3 · Real per capita capital growth. Other explanatory variables included in each of the nine regression results reported above: logarithm of initial income, logarithm of initial secondary school enrollment, ratio of government consumption expenditures to GDP, inflation rate, and ratio of exports plus imports to GDP. Notes: King and Levine (1993b) define 2 percent growth as 0.02. For comparability with subsequent tables, we have redefined 2 percent growth as 2.00 and adjusted the coefficients by a factor of 100.
do not consider what actually causes the change in financial development. They simply illustrate the potentially large long-term growth effects from changes in financial development. La Porta, Lopez-de-Silanes and Shleifer (2002) use an alternative indicator of financial development. They examine the degree of public ownership of banks around the world. To the extent that publicly-owned banks are less effective at acquiring information about firms, exerting corporate governance, mobilizing savings, managing risk, and facilitating transactions, then this measure provides direct evidence on connection between economic growth and the services provided by financial intermediaries. The authors show that (1) higher degrees of public ownership are associated with lower levels of bank development and (2) high levels of public ownership of banks are associated with slower economic growth. While addressing many of the weaknesses in earlier work, cross-country growth regressions do not eliminate them. Thus, while KL show that finance predicts growth, they do not deal formally with the issue of causality [Shan, Morris and Sun (2001)]. While researchers improve upon past measures of financial development, they only focus on
Ch. 12:
Finance and Growth: Theory and Evidence
893
Table 2 Growth and initial financial depth, 1960–1989 Dependent variable Real per capita GDP growth, 1960–1989 R2 Real per capita capital growth, 1960–1989 R2 Productivity growth, 1960–1989 R2
DEPTH in 1960 2.8∗∗ (0.001) 0.61 1.9∗∗ (0.001) 0.63 2.2∗∗ (0.001) 0.58
Sources: King and Levine (1993b, Table VIII) and Levine (1997, Table 3). ∗ Significant at the 0.10 level. ∗∗ Significant at the 0.05 level. (p-values in parentheses.) Observations: 57. Variable definitions: DEPTH = Liquid liabilities/GDP, Productivity growth = Real per capita GDP growth− 0.3 · Real per capita capital growth. Other explanatory variables included in each of the regression results reported above: logarithm of initial income, logarithm of initial secondary school enrollment, ratio of government consumption expenditures to GDP, inflation rate, and ratio of exports plus imports to GDP. Notes: King and Levine (1993b) and Levine (1997) define 2 percent growth as 0.02. For comparability with subsequent tables, we have redefined 2 percent growth as 2.00 and adjusted the coefficients by a factor of 100.
one segment of the financial system, banks, and their indicators do not directly measure the degree to which financial systems ameliorate information and transaction costs. 3.1.3. Adding stock markets to cross-country studies of growth There are good reasons to study the relationship between long-run economic growth and the operation of equity markets. First, as stressed above, theoretical debate exits on whether larger, more liquid equity markets exert a positive or negative influence on economic growth, capital accumulation, and productivity growth. Second, as stressed above, some theories focus on the competing roles of banks and markets in funding corporate expansion, while others stress that banks and markets may arise, coexist, and prosper by providing different financial functions to the economy, and still other theories stress complementarities between banks and markets. Thus, simultaneously considering the potential roles of banks and markets permits one to distinguish among competing theories and provide evidence to policy makers on the independent roles of markets and banks in the process of economic growth. Levine and Zervos (1998a, henceforth LZ) construct numerous measures of stock market development to assess the relationship between stock market development and
894
R. Levine Table 3 Stock market and bank development predict growth, 1976–1993
Dependent variable (1976–1993)
Real per capita GDP growth Real per capita capital growth Productivity growth
Independent variables (1976) Bank credit
Turnover
R2
1.31∗∗ (0.022) 1.48∗∗ (0.025) 1.11∗∗ (0.020)
2.69∗∗ (0.005) 2.22∗∗ (0.024) 2.01∗∗ (0.029)
0.50 0.51 0.40
Source: Levine and Zervos (1998a, 1998b, Table 3). ∗ Significant at the 0.10 level. ∗∗ Significant at the 0.05 level. (p-values in parentheses.) Observations: 42 for the real per capita GDP growth regression and 41 for the others. Variable definitions: Bank credit = Bank credit to the private sector/GDP in 1976 or the closest date with data, Turnover = Value of the trades of domestic shares on domestic exchanges as a share of market capitalization of domestic shares in 1976 or the closest date with data, Productivity growth = Real per capita GDP growth− 0.3 · Real per capita capital growth. Other explanatory variables included in each of regression results reported above: logarithm of initial income, logarithm of initial secondary school enrollment, ratio of government consumption expenditures to GDP, inflation rate, black market exchange rate premium, and frequency of revolutions and coups. Notes: Levine and Zervos define 2 percent growth as 0.02. For comparability with subsequent tables, we have redefined 2 percent growth as 2.00 and adjusted the coefficients by a factor of 100.
economic growth, capital accumulation, and productivity growth in a sample of 42 countries over the period 1976–93.12 They control for many other potential growth determinants, including banking sector development. Their study builds on pioneering work by Atje and Jovanovic (1993). For brevity, I focus on only one of LZ’s liquidity indicators, the turnover ratio. This equals the total value of shares traded on a country’s stock exchanges divided by stock market capitalization (the value of listed shares on the country’s exchanges). The turnover ratio is not a direct measure of trading costs or of the ability to sell securities at posted prices. Rather, the turnover ratio measures trading relative to the size of the market. It therefore reflects trading frictions and information that induces transactions. This ratio exhibits substantial cross-country variability. Very active markets such as Japan and the United States had turnover ratios of almost 0.5 during the 1976–93 period, while less liquid markets, such as Bangladesh, Chile, and Egypt have turnover ratios of 0.06 or less. As summarized in Table 3, LZ find that the initial level of stock market liquidity and the initial level of banking development (Bank Credit) are positively and significantly 12 These measures build on Demirgüç-Kunt and Levine (1996).
Ch. 12:
Finance and Growth: Theory and Evidence
895
correlated with future rates of economic growth, capital accumulation, and productivity growth over the next 18 years even after controlling for initial income, schooling, inflation, government spending, the black market exchange rate premium, and political stability. Bank credit equals bank credit to the private sector as a share of GDP.13 These results are consistent with the view that stock market liquidity facilitates longrun growth [Levine (1991), Holmstrom and Tirole (1993), Bencivenga, Smith and Starr (1995)], but inconsistent with models that emphasize the negative aspects of stock markets liquidity [Bhide (1993)]. Furthermore, the results do not lend much support to models that emphasize the tensions between bank-based and market-based systems. Rather, the results suggest that stock markets provide different financial functions from those provided by banks, or else they would not both enter the growth regression significantly. The sizes of the coefficients also suggest an economically meaningful relationship. For example, the estimated coefficient implies that a one-standard-deviation increase in initial stock market liquidity (0.30) would increase per capita GDP growth by 0.80 percentage points per year (2.7 · 0.3). Accumulating over 18 years, this implies real GDP per capita would have been over 15 percentage points higher by the end of the sample. Similarly, the estimated coefficient on Bank Credit implies a correspondingly large growth effect. That is, a one-standard deviation increase in Bank Credit (0.5) would increase growth by 0.7 percentage point per year (1.3 · 0.5). Taken together, the results imply that if a country had increased both stock market liquidity and bank development by one-standard deviation, then by the end of the 18-year sample period, real per capita GDP would have been almost 30 percent higher and productivity would have been almost 25 percent higher. As emphasized throughout, these conceptual experiments do not consider the underlying causes of the change in the operation of the financial sector. The examples simply illustrate the potential growth effects of financial development. LZ go onto argue that the link between stock markets, banks, and growth runs most robustly through productivity growth, rather than physical capital accumulation, which is consistent with some theoretical models [Levine (1991), Bencivenga, Smith and Starr (1995)]. LZ also find that stock market size, as measured by market capitalization divided by GDP, is not robustly correlated with growth, capital accumulation, and productivity improvements. This is consistent with theory. Simply listing on the national stock exchange does not necessarily foster resource allocation. Rather, it is the ability to trade the economy’s productive technologies easily that influences resource allocation and growth. There are a number of weaknesses, however, associated with the LZ approach. First, while they show that stock market liquidity and bank development predict economic growth, they do not deal formally with the issue of causality.
13 Note, King and Levine’s (1993a, 1993b, 1993c) PRIVY measures total credit flowing to the private sector,
while Levine and Zervos’s (1998a, 1998b) Bank Credit measures credit by banks to the private sector.
896
R. Levine
Second, there are difficulties in measuring liquidity as discussed by Grossman and Miller (1988). LZ do not measure the direct costs of conducting equity transactions. Furthermore, they do not control for the possibility that the arrival of information and the processing of that information may differ across countries and thereby induce crosscountry differences in trading that does not reflect liquidity as defined by theory. While LZ confirm their results using three additional measures of liquidity, measurement issues remain.14 Third, more broadly, the liquidity indicators measure domestic stock transactions on a country’s national stock exchanges. The physical location of the stock market, however, may not necessarily matter for the provision of liquidity unless there are impediments to cross-location transactions. Physical location will matter less – and this measurement problem will matter more – if economies become more financially integrated. Guiso, Sapienza and Zingales (2002), however, find that local financial conditions matter even in a single country – Italy. They show that local financial conditions influence economic performance across the different regions of Italy. That is, local financial development is an important determinant of the economic success of an area even within a single country. Their results suggest that international financial integration is unlikely to eliminate the importance of national financial systems in the near future.15 Fourth, even more generally, the link between trading and future economic growth may not represent a link between liquidity and growth as suggested by some theories [Levine (1991), Bencivenga, Smith and Starr (1995)]. The liquidity-stock market link may be generated by a third factor that produces both a surge in trading and a subsequent acceleration in economic growth, but where trading does not induce the growth acceleration. For instance, positive news about a technology shock may elicit different opinions about which sectors and firms will benefit most from the innovation. This would produce lots of trading today because of these differences of opinion. The subsequently surge in economic growth is due to the positive technology shock, not the increase in stock transactions. In this “model”, trading does not necessarily facilitate the ability of the economy to exploit the growth benefits of the technology shock. From this perspective, it is difficult to interpret the LZ results as implying that liquidity fosters economic growth. Fifth, while LZ include measures of the functioning of stock markets and banks, they exclude other components of the financial sector, e.g., bond markets and the financial
14 LZ examine three additional measures of liquidity. First, the value traded ratio equals the total value of
domestic stocks traded on domestic exchanges as a share of GDP. This measures trading relative to the size of the economy. The next two measures of liquidity measure trading relative to stock price movements: (1) the value traded ratio divided by stock return volatility, and (2) the turnover ratio divided by stock return volatility. 15 Levine and Schmukler (2003, 2004) find that international cross-listing by emerging market firms can hurt the operation of the emerging market itself with potentially adverse implications for economic development according to the conclusions in Guiso, Sapienza and Zingales (2002). In terms of international banking, Levine (2004) finds that regulatory restrictions on foreign bank entry hurt the efficiency of domestic banking sector operations.
Ch. 12:
Finance and Growth: Theory and Evidence
897
services provided by nonfinancial firms. Beck, Demirgüç-Kunt and Levine (2001) show that in many countries private bond market capitalization is more than half the capitalization of national equity markets and public bond markets are frequently larger than stock markets. Furthermore, over the period 1980–1995, new issuances of private bonds were greater than public offerings of stock in many countries. Fink, Haiss and Hristoforova (2003) examine the impact of bond market development on real output in 13 highly developed economies over the period 1950–2000. Using Granger causality tests and co-integration methods, the bulk of their evidence indicates that bond market development influences real economic activity. Furthermore, Beck, Demirgüç-Kunt and Levine (2001) show that life insurance and private pension fund assets rival banks in some countries, while Berger, Hasan and Klapper (2005) indicate that small, community banks boost growth in many developing countries. Thus, more work remains on incorporating bond markets and nonbank institutions into finance-growth literature. Sixth, stock markets may do more than provide liquidity. Stock markets may provide mechanisms for hedging and trading the idiosyncratic risk associated with individual projects, firms, industries, sectors, and countries. While a vast literature examines the pricing of risk, there exists very little empirical evidence that directly links risk diversification services with long-run economic growth. While LZ do not find a strong link between economic growth and the ability of investors to diversify risk internationally, they have extremely limited data on international integration. Future work needs to more fully assess the links between stock markets, banks, and economic growth. 3.1.4. Using instrumental variables in cross-country studies of growth While KL and LZ show that financial development predicts economic growth, these results do not settle the issue of causality. It may simply be the case that financial markets develop in anticipation of future economic activity. Thus, finance may be a leading indicator rather than a fundamental cause. To assess whether the finance-growth relationship is driven by simultaneity bias, one needs instrumental variables that explain cross-country differences in financial development but are uncorrelated with economic growth beyond their link with financial development and other growth determinants. Levine (1998, 1999) and Levine, Loayza and Beck (2000) use the La Porta et al. (1998, henceforth LLSV) measures of legal origin as instrumental variables. In particular, LLSV (1998) show that legal origin – whether a country’s Commercial/Company law derives from British, French, German, or Scandinavian law – importantly shapes national approaches to laws concerning creditors and the efficiency with which those laws are enforced. Since finance is based on contracts, legal origins that produce laws that protect the rights of external investors and enforce those rights effectively will do a correspondingly better job at promoting financial development.16 Indeed, LLSV (1998), Levine (1998, 1999, 2003), and 16 In terms of identifying why legal tradition influences the operation of the financial system, see Beck,
Demirgüç-Kunt and Levine (2003b, 2005a).
898
R. Levine
Levine, Loayza and Beck (2000) trace the effect of legal origin to laws and enforcement and then to financial development. Since most countries obtained their legal systems through occupation and colonization, the legal origin variables may be plausibly treated as exogenous. Following Levine, Loayza and Beck (2000, henceforth LLB) analysis of 71 countries, consider the generalized method of moments (GMM) regression: G(j ) = α + βF (i) + γ X + ε. G(j ) is real per capita GDP growth over the 1960–95 period. The legal origin indicators, Z, are used as instrumental variables for the measures of financial development, F (i). X is treated as an included exogenous variable. LLB use linear moment conditions, which amounts to the requirement that the instrumental variables (Z) be uncorrelated with the error term (ε). The economic meaning of these conditions is that legal origin may affect per capita GDP growth only through the financial development indicators and the variables in the conditioning information set, X. LLB extend the King and Levine (1993a, 1993b) measures of financial intermediary development through to 1995, improve the deflating of the financial development indicators, and add a new measure of overall financial development.17 The new measure of financial development, Private Credit, equals the value of credits by financial intermediaries to the private sector divided by GDP. The measure isolates credit issued to the private sector and therefore excludes credit issued to governments, government agencies, and public enterprises. Also, it excludes credits issued by central banks. Unlike the LZ Bank Credit measures, Private Credit included credits issued by non-deposit money bank. Not surprisingly, there is enormous cross-country variation in Private Credit. Private Credit is less than 10 percent of GDP in Zaire, Sierra Leone, Ghana, Haiti, and Syria, while it is greater than 85 percent of GDP in Switzerland, Japan, the United States, Sweden, and the Netherlands. The LLB results indicate a very strong connection between the exogenous component of financial intermediary development and long-run economic growth. They use various measures of financial intermediary development and different conditioning information sets, i.e., different X’s. They find that the exogenous component of financial development is closely tied to long-run rates of per capita GDP growth. Furthermore, the data do not reject the test of the over-identifying restrictions. The inability to reject the orthogonality conditions plus the finding that the legal origin instruments (Z) are highly correlated with financial intermediary development indicators (i.e., the null hypothesis that the legal origin variables does not explain the financial intermediary indicators 17 LLB (2000) improves upon past measures of financial intermediary development by more accurately
deflating nominal measures of financial intermediary liabilities and assets. Specifically, while financial intermediary balance sheet items are measured at the end of the year, GDP is measured over the year. LLB deflate end-of-year financial balance sheet items by end of year consumer price indices (CPI) and deflate the GDP series by the annual CPI. Then, they compute the average of the real financial balance sheet item in year t and t − 1 and divide this average by real GDP measured in year t.
Ch. 12:
Finance and Growth: Theory and Evidence
899
is rejected at the 0.01 significance level), suggest that the instruments are appropriate. These results indicate that the strong link between financial development and growth is not due to simultaneity bias. The estimated coefficient can be interpreted as the effect of the exogenous component of financial intermediary development on growth. LLB’s (2000) instrumental variable results also indicate an economically large impact of financial development on growth. For example, India’s value of Private Credit over the period 1960–95 was 19.5 percent of GDP, while the mean value for developing countries was 25 percent of GDP. The estimated coefficients in LLB suggest that an exogenous improvement in Private Credit in India that had pushed it to the sample mean for developing countries would have accelerated real per capita GDP growth by an additional 0.6 of a percentage point per year.18 Similarly, if Argentina had moved from its value of Private Credit (16) to the developing country sample mean, it would have grown more than one percentage point faster per year. This is large considering that growth only averaged about 1.8 percent per year over this period. As emphasized throughout, however, these types of conceptual experiments must be treated as illustrative because they do not account for how to increase financial intermediary development. While LLB interpret their results as implying that financial development boosts steady-state growth, Aghion, Howitt and Mayer-Foulkes (2005) challenge that conclusion. They first develop a model of technological change that predicts that countries with levels of financial development above a critical, threshold level will converge in growth rates. Among these countries, financial development positively influences the rate of convergence, so the financial development exerts positive but diminishing influence on steady-state levels of real per capita output. They find empirical support for the model’s predictions. Financial development explains (i) whether there is convergence or not, and (ii) the rate of convergence (when there is convergence), but Aghion, Howitt and Mayer-Foulkes (2005) find that financial development does not exert a direct effect on steady-state growth. 3.2. Panel, time-series, and case-studies of finance and growth Studies of finance and growth have also employed panel data techniques, pure timeseries methodologies, and case-studies to ameliorate a number of statistical problems with pure cross-country investigations. This section discusses the panel approach in some depth and finishes with shorter discussions of pure time-series and case-study approaches.
18 To get this, note that LLB take logarithms of the financial intermediary indicators to reduce the effect of
outliers, so that the change in financial development is ln(25) − ln(19.5) = 0.25. Then, use their smallest parameter estimate on Private Credit from their Table 3, which equals 2.5. Thus, the acceleration in growth is given by 2.5 · (0.25) = 0.63.
900
R. Levine
3.2.1. The dynamic panel methodology LLB (2000) and Beck, Levine and Loayza (2000, henceforth BLL) use a panel GMM estimator that improves upon pure cross-country work in three respects [Arellano and Bond (1991)]. The regression equation in levels can be specified in the following form: 1 2 yi,t = α Xi,t−1 + β Xi,t + µi + λt + εi,t
(1)
where y represents the dependent variable, X 1 represents a set of lagged explanatory variables and X 2 a set of contemporaneous explanatory variables, µ is an unobserved country-specific effect, λ is a time-specific effect, ε is the time-varying error term, and i and t represent country and (5-year) time period, respectively. The first benefit from moving to a panel is the ability to exploit the time-series and cross-sectional variation in the data. LLB construct a panel that consists of data for 77 countries over the period 1960–95. The data are averaged over seven non-overlapping five-year periods. Moving to a panel incorporates the variability of the time-series dimension. Specifically, the within-country standard deviation of Private Credit is 15%, which in the panel estimation is added to the between-country standard deviation of 28%. Similarly, for real per capita GDP growth, the within-country standard deviation is 2.4% and the between-country standard deviation is 1.7%.19 This also raises a potential disadvantage from moving to panel data. With panel data, we employ data averaged over five-year periods, yet the models we are using to interpret the data are typically models of steady-state growth. To the extent that five years does not adequately proxy for long-run relationships, the panel methods may imprecisely assess the finance growth link. The second benefit from moving to a panel is that it avoids biases associated with cross-country regressions: With cross-country regressions, the unobserved countryspecific effect is part of the error term so that correlation between m and the explanatory variables results in biased coefficient estimates. Furthermore, if the lagged dependent variable is included in X 1 (which is the norm in cross-country regressions), then the country-specific effect is certainly correlated with X 1 . First differencing the regression equation eliminates the country-specific effect. 1 2 1 2 yi,t − yi,t−1 = α Xi,t−1 (2) + β Xi,t + (εi,t − εi,t−1 ). − Xi,t−2 − Xi,t−1 This, however, introduces correlation between the new error term εi,t − εi,t−1 and the 1 1 lagged dependent variable yi,t−1 − yi,t−2 when it is included in Xi,t−1 − Xi,t−2 . One can use lagged values of the explanatory variables in levels as instruments. Assuming (i) no serial correlation and (ii) the explanatory variables X (X = [X 1 X 2 ]) are weakly
19 The within-country standard deviation is calculated using the deviations from country averages, whereas
the between-country standard deviation is calculated from the country averages.
Ch. 12:
Finance and Growth: Theory and Evidence
901
exogenous, the following moment conditions hold. E Xi,t−s (εi,t − εi,t−1 ) = 0 for s ≥ 2; t = 3, . . . , T .
(3)
This difference estimator consists of the regression in differences plus Equation (3). The third benefit from moving to a panel is that it permits the use of instrumental variables for all regressors and thereby provides more precise estimates of the financegrowth relationship. As discussed, researchers use legal origin instruments to extract the exogenous component of financial development. These pure cross-sectional estimators, however, do not control for the endogeneity of all the other explanatory variables. This can lead to inappropriate inferences on the coefficient on financial development. Building on this difference panel estimator, Arellano and Bover (1995) propose a system estimator that jointly estimates the regression in levels (Equation (1)) and the equation in differences (Equation (2)) in order to (i) re-incorporate the cross-country variation from the levels regression and (ii) reduce the likelihood that weak instruments bias the estimated coefficients and standard errors. 3.2.2. Dynamic panel results on financial intermediation and growth LLB use the system estimator to examine the relationship between financial intermediary development and growth, while BLL examine the relationship between financial development and the sources of growth, i.e., productivity growth, physical capital accumulation, and savings. They examine an assortment of indicators of financial intermediary development and also use a variety of conditioning information sets to assess the robustness of the results [Levine and Renelt (1992)]. Here, we summarize the results in Table 4 using the Private Credit measure of financial development described above and a simple set of control variables. The results indicate a positive relationship between the exogenous component of financial development and economic growth, productivity growth, and capital accumulation. The regressions pass the standard specification tests. Table 4 presents both (1) instrumental variable results using a pure a cross-sectional analysis where the legal origin variables are the instruments and (2) the dynamic panel results just described. Remarkably the coefficient estimates are very similar using the two procedures and economically significant. Thus, the large, positive relationship between economic growth and Private Credit does not appear to be driven by simultaneity bias, omitted countryspecific effects, or the routine use of lagged dependent variables in cross-country growth regressions. While BLL go on to argue that the finance-capital accumulation link is not robust to alternative specifications, they demonstrate a robust link between financial development indicators and both economic growth and productivity growth. The regression coefficients suggest an economically large impact of financial development on economic growth. For example, Mexico’s value for Private Credit over the period 1960–95 was 22.9% of GDP. An exogenous increase in Private Credit that had
902
R. Levine
Table 4 Growth, Productivity growth, and Capital accumulation, panel GMM and OLS, 1960–1995 1. Dependent variable: Real per capita GDP growth Estimation procedure PRIVATE Countries Obs. CREDIT IV-cross-country GMM-panel
2.22∗∗ (0.003) 2.40∗∗ (0.001)
63
63
77
365
2. Dependent variable: Productivity growth Estimation procedure PRIVATE Countries CREDIT IV-cross-country GMM-panel
1.50∗∗ (0.004) 1.33∗∗ (0.001)
GMM-panel
2.83∗∗ (0.006) 3.44∗∗ (0.001)
OIR-test1
63
63
2.036
77
365
63
63
77
365
Sargan test2 (p-value)
Serial correlation test3 (p-value)
0.183
0.516
Sargan test2 (p-value)
Serial correlation test3 (p-value)
0.205
0.772
Sargan test2 (p-value)
Serial correlation test3 (p-value)
0.166
0.014
0.577
Obs.
3. Dependent variable: Capital per capita growth Estimation procedure PRIVATE Countries Obs. CREDIT IV-cross-country
OIR-test1
OIR-test1
6.750
Source: Beck, Levine and Loayza (2000). ∗ Significant at the 0.10 level. ∗∗ Significant at the 0.05 level. (p-values in parentheses.) IV-cross-country: Cross-country instrumental variables with legal origin as instruments, estimated using GMM. GMM-panel: Dynamic panel (5-year averages) generalized method of moments using system estimator. Other explanatory variables: logarithm of initial income per capita, average years of schooling. PRIVATE CREDIT: Logarithm(credit by deposit money banks and other financial institutions to the private sector divided by GDP). 1 The null hypothesis is that the instruments used are not correlated with the residuals from the respective regression. Critical values for OIR-Test (2 d.f.): 10% = 4.61; 5% = 5.99. 2 The null hypothesis is that the instruments used are not correlated with the residuals from the respective regression. 3 The null hypothesis is that the errors in the first-difference regression exhibit no second-order serial correlation.
Ch. 12:
Finance and Growth: Theory and Evidence
903
brought it up to the sample median of 27.5% would have resulted in a 0.4 percentage point higher real per capita GDP growth per year.20 While BLL and LLB examine linear models, recent research suggests that the impact of financial development on capital accumulation, productivity growth, and overall real per capita GDP growth may depend importantly on other factors. Using the same econometric methods and data, Rioja and Valev (2004a) find that finance boosts growth in rich countries primarily by speeding-up productivity growth, while finance encourages growth in poorer countries primarily by accelerating capital accumulation. Furthermore, Rioja and Valev (2004b) find that the impact may be nonlinear. They find that countries with very low levels of financial development experience very little growth acceleration from a marginal increase in financial development, while the affect is larger for rich countries and particular large for middle-income countries. It would be nice to know, however, what produces these nonlinearities. Finally, Rousseau and Wachtel (2002) show that the positive impact of financial development on growth diminishes with higher rates of inflation. Emphasizing that not all indicators of financial development measure the same forces, Benhabib and Spiegel (2000) examine the relationship between an assortment of financial intermediary development indicators and economic growth, investment, and total factor productivity growth. While they use a panel estimator, they do not use the system estimator described above that allows for the endogeneity of all the regressors and the routine use of lagged dependent variables. They find that the indicators of financial development are correlated with both total factor productivity growth and the accumulation of both physical and human capital. Their paper raises an important qualification, however. Different indicators of financial development are linked with different components of growth (total factor productivity, physical capital accumulation, and human capital accumulation). Their findings reiterate an important qualification running throughout this survey: it is difficult to measure financial development and link empirical constructs with theoretical concepts. Loayza and Ranciere (2002) extend this line of empirical inquiry by differentiating between the long-run and short-run relationships connecting finance and economic activity. They note that short-run surges in bank lending can actually signal the onset of financial crises and economic stagnation. They stress that it is therefore crucial to consider simultaneously the short-run and long-run effects of financial development. For instance, while finance is positively associated with economic growth in a broad cross-section of countries, this relationship does not hold in Latin America, which has been subject to severe and repeated banking crises. Using a panel, Loayza and Ranciere (2002) estimate an encompassing model of long-run and short-run effects. Using the LLB measure of financial intermediary development (Private Credit), they find that a
20 This results follows from ln(27.5) − ln(22.9) = 0.18 and 0.18 · 2.4 = 0.43, where 2.4 is the parameter
estimate from the panel regression.
904
R. Levine Table 5 Stock markets, banks, and growth: Panel GMM and OLS, 1975–1998
Dependent variable: Real per capita GDP growth Estimation procedure Bank credit Turnover Countries
OLS-cross-country GMM-panel
1.47∗∗ (0.001) 1.76∗∗ (0.001)
0.79∗∗ (0.025) 0.96∗∗ (0.001)
Obs.
Sargan test1 (p-value)
Serial correlation test2 (p-value)
146
0.488
0.60
40 40
Source: Beck and Levine (2004, Tables 2 and 3). ∗ Significant at the 0.10 level. ∗∗ Significant at the 0.05 level. (p-values in parentheses.) OLS: Ordinary Least Squares with heteroscedasticity consistent standard errors. GMM: Dynamic panel Generalized Method of Moments using system estimator. Bank credit = logarithm(credit by deposit money banks to the private sector as a share of GDP). Turnover = logarithm(value of the trades of domestic shares on domestic exchanges as a share of market capitalization of domestic shares). Other explanatory variables included in each of the regression results reported above: logarithm of initial income and logarithm of initial secondary school enrollment. 1 The null hypothesis is that the instruments used are not correlated with the residuals. 2 The null hypothesis is that the errors in the first-difference regression exhibit no second-order serial correlation.
positive long-run relationship between financial development and growth co-exists with a generally negative short-run link.21 3.2.3. Dynamic panel results and stock market and bank development Rousseau and Wachtel (2000) examine the relationship between stock markets, banks, and growth, using annual data and the difference estimator. Beck and Levine (2004) use data averaged over five-year periods to focus on longer-run growth factors, use the system estimator to mitigate potential biases associated with the difference estimator, and extend the sample through 1998 (from 1995).22 Table 5 indicates that the exogenous component of both stock market development and bank development help predict economic growth. As shown, the coefficient estimates from the two methods are very similar. The panel procedure passes the standard 21 For more on distinguishing the short-run and long-run effects of financial development, see Fisman and
Love (2003b). 22 There are additional econometric problems created when studying stock markets, banks, and economic
growth. There are many fewer countries and years when incorporating stock markets, which can lead to overfitting of the data and potential mis-leading inferences. Beck and Levine (2004) describe and use variants of the dynamic panel estimator to reduce the likelihood that over-fitting biases the results.
Ch. 12:
Finance and Growth: Theory and Evidence
905
specification tests, which increases confidence in the assumptions underlying the econometric methodology. While not shown, stock market capitalization is not closely associated with growth. Thus, it is not listing per se that is important for growth; rather, it is the ability of agents to exchange ownership claims on an economy’s productive technologies that is relevant for economic growth. Table 5 estimates are economically meaningful and consistent with magnitudes obtained using different methods. If Mexico’s Turnover Ratio had been at the average of the OECD countries (68%) instead of the actual 36% during the period 1996–98, it would have grown 0.6 percentage points faster per year. Similarly, if its Bank Credit had been at the average of all OECD countries (71%) instead of the actual 16%, it would have grown 2.6 percentage points faster per year. These results suggest that the exogenous components of both bank and stock market development have an economically large impact on economic growth. 3.2.4. Time series studies A substantial time-series literature examines the finance-growth relationship using a variety of time-series techniques. These studies frequently use Granger-type causality tests and vector autoregressive (VAR) procedures to examine the nature of the financegrowth relationship [e.g., Arestis and Demetriades (1997)]. Research has progressed by using better measures of financial development, employing more powerful econometric techniques, and by examining individual countries in much greater depth. Some initial time-series studies emphasize the importance of measuring financial development accurately, suggesting that studies that use more precise measures of financial development tend to find a growth-enhancing impact of financial development. Jung (1986) and Demetriades and Hussein (1996) use measures of financial development such as the ratio of money to GDP. They find the direction of causality frequently runs both ways, especially for developing economies. However, Neusser and Kugler (1998) use measures of the value-added provided by the financial system instead of simple measures of the size of the financial system. They find that finance boosts growth. Furthermore, Rousseau and Wachtel (1998) conduct time-series tests of financial development and growth for five countries over the past century using measures of financial development that include the assets of both banks and non-banks. They document that the dominant direction of causality runs from financial development to economic growth. Finally, Arestis, Demetriades and Luintel (2001) augment time-series studies of finance and growth by using measures of both stock market and bank development. They find additional support for the view that finance stimulates growth but raise some cautions on the size of the relationship. They use quarterly data and apply time series methods to five developed economies and show that while both banking sector and stock market development explain subsequent growth, the effect of banking sector development is substantially larger than that of stock market development. The sample size, however, is very limited and it is not clear whether the use of quarterly data and a
906
R. Levine
vector error correction model fully abstract from high frequency factors influencing the stock market, bank, and growth nexus to focus on long-run economic growth. Additional econometric sophistication has also been brought to bear on the finance and growth question. In a broad study of 41 countries over the 1960–1993, Xu (2000) uses a VAR approach that permits the identification of the long-term cumulative effects of finance on growth by allowing for dynamic interactions among the explanatory variables.23 Xu (2000) rejects the hypothesis that finance simply follows growth. Rather, the analyses indicate that financial development is important for long-run growth. More recently, Christopoulos and Tsionas (2004) note that many time-series studies yield unreliable results due to the short time spans of typical data sets. Thus, they use panel unit root tests and panel cointegration analyses to examine the relationship between financial development and economic growth in ten developing countries to yield causality inferences within a panel context that increases sample size. In contrast to Demetriades and Hussein (1996), Christopoulos and Tsionas (2004) find strong evidence in favor of the hypothesis that long-run causality runs from financial development to growth and that there is no evidence of bi-directional causality. Furthermore, they find a unique cointegrating vector between growth and financial development, and emphasize the long-run nature of the relationship between finance and growth. There has also been a movement away from applying time-series methods to a variety of countries and toward examining individual countries, which allows research to design country-specific measures of financial development and expand the time-series dimension of the analyses in some cases. Rousseau and Sylla (1999) expand Rousseau’s (1998) examination of the historical role of finance in U.S. economic growth to include stock markets. They use a set of multivariate time-series models that relate measures of banking and equity market activity to investment, imports, and business incorporations over the 1790–1850 period. Rousseau and Sylla (1999) find strong support for the theory of “finance led growth” in United States. Moving beyond the U.S., Rousseau and Sylla (2001) study seventeen countries over the period 1850–1997. They also find evidence consistent with the view the financial development stimulated economic growth in these economies. In a study of the Meiji period in Japan (1868–1884), Rousseau (1999) uses a variety of VAR procedures and concludes that the financial sector was instrumental in promoting Japan’s explosive growth prior to the First World War. In a different study, Rousseau (1998) examines the impact of financial innovation in the U.S. on financial depth over the period 1872–1929. Innovation is proxied by reductions in the loan-deposit spread. The impact on the size of the financial intermediary sector is assessed using unobservable components methods. The paper finds that permanent reductions of 1% in the spread of New York banks are associated with increases in financial depth that range from 1.7% to nearly 4%. While not a direct link to growth, these findings develop a direct link running from financial innovation to increases in financial depth, which is commonly associated with economic growth in other studies. 23 In a narrower study, Luintel and Khan (1999) find some evidence of bi-directional causality between
finance and growth in VAR analysis of developing countries.
Ch. 12:
Finance and Growth: Theory and Evidence
907
Bekaert, Harvey and Lundblad (2001, 2005) examine the effects of opening equity markets to foreign participation.24 One statistical innovation in their work is the use of over-lapping data. Many time-series studies use annual observations and even quarterly data to maximize the information included their analyses. Bekaert, Harvey and Lundblad (2005), however, use data averaged over five-year periods to focus on growth rather than higher frequency relationships, but they use over-lapping data to avoid the loss of information inherent in using non-over-lapping data. Specifically, one observation includes data averaged from 1990–1995 and the next period includes data averaged from 1991–1996. They adjust the standard errors accordingly and conduct an array of sensitivity checks, though the procedure does not formally deal with simultaneity bias. Consistent with Levine and Zervos (1998a), Bekaert, Harvey and Lundblad (2001, 2005) show that financial liberalization boosts economic growth by improving the allocation of resources and the investment rate. 3.2.5. Novel case-studies Jayaratne and Strahan (1996) undertake a fascinating examination of the impact of finance on economic growth by examining individual states of the United States. Since the early 1970s, 35 states relaxed impediments on intrastate branching. They estimate the change in economic growth rates after branch reform relative to a control group of states that did not reform. They use a pooled time-series, cross-sectional dataset to assess the impact of liberalizing branching restrictions on state growth. Jayaratne and Strahan (1996) show that branch reform boosted bank-lending quality and accelerated real per capita growth rates, while Dehejia and Lleras-Muney (2003) confirm and extend these findings by also examining the impact of deposit insurance. By comparing states within the United States, the paper eliminates problems associated with country-specific factors. The paper also uses a natural identifying condition, the change in branching restrictions, to trace through the impact of financial development on economic growth. Importantly, the paper finds little evidence that branch reform boosted lending. Rather, branch reform accelerated economic growth by improving the quality of bank loans and the efficiency of capital allocation.25 Some issues remain, however. While Jayaratne and Strahan (1996) control for state investment and tax receipts, it is difficult to control fully for other factors influencing growth in the individual states. Similarly, while the authors show that (i) there is no correlation between the business cycle and the timing of regulation and (ii) deregulation does not forecast a boom in lending, it is difficult to rule out the possibility that states liberalize banking due to 24 For further analyses on the growth effects of international financial liberalization, see Henry (2000, 2003),
Levine and Zervos (1998b), Edison et al. (2002), and Klein and Olivei (2001) and the references therein. 25 Note, Jayaratne and Strahan (1998) show that with bank deregulation, better-managed, lower costs banks
expand at the expense of inefficient banks. On an international level, Demirgüç-Kunt, Laeven and Levine (2004) show that regulatory restrictions reduce banking sector efficiency and Beck, Demirgüç-Kunt and Levine (2003d) find that regulatory restrictions on bank competition tend to increase the fragility of banks.
908
R. Levine
expected growth-enhancing structural changes in the economy that do require more lending but better lending. Dehejia and Lleras-Muney (2003) also examine the growth experiences of states across the U.S. They too find that financial development boosts growth, but they also show that deposit insurance frequently induced indiscriminate credit expansions with adverse effects on growth. Again, the results suggest that it is the quality, not the quantity, of lending that matters. In sum, these innovative studies provide empirical support for the view that well-functioning banks improve the allocation of capital and hence economic growth. In terms of the early years of the United States, Wright (2002) provides a lucid and detailed examination of how the U.S. financial system drove America’s transformation after 1780 from an agricultural economy to a thriving industrial power. The book’s core thesis is that “. . . the U.S. financial system created the conditions necessary for the sustained domestic economic growth . . . that scholars know occurred in the nineteenth century”. Most impressively, Wright’s (2002) research is filled with specific examples of the emergence of new financial arrangement to facilitate the acquisition of information about firms (pp. 26–50), to monitor managers and to align the interests of creditors and firm insiders (pp. 37–41), and to facilitate the trading, hedging, and pooling of risk (pp. 51–75). For example, in response to principal-agent problems, U.S. corporations in the 18th century increasingly forced managers to hold large quantities of stock in the corporation to align their personal financial interests with those of the firm (p. 39). As another example, after suffering through high default rates, U.S. bankers quickly learned to monitor borrowers more carefully by continuously reviewing the cash-flows of borrowers to identify unusual activity, forcing debtors to report their actions at regular board meetings and granting additional privileges only to debtors demonstrating good behavior, and forcing borrowers to allocate funds toward very specific investments along with other very restrictive covenants (pp. 34–35). While the book does not provide formal statistical evidence that financial development accelerated economic growth in the early decades after U.S. independence, Wright (2002) make a different, distinguishing contribution: He documents how specific financial contracts, markets, and institutions arose to ease information and transactions costs and hence influence the resource allocation decisions of a country. Guiso, Sapienza and Zingales (2002) examine the individual regions of Italy. Using an extraordinary dataset on households and financial services across Italy, they examine the effects of differences in local financial development on economic activity across the regions of Italy. Guiso, Sapienza and Zingales (2002) find that local financial development (i) enhances the probability that an individual starts a business, (ii) increases industrial competition, and (iii) promotes the growth of firms. These results are weaker for large firms, which can more easily raise funds outside of the local area. This study ameliorates many of the weaknesses associated with examining growth across countries. Consider also Haber’s (1991, 1997) impressive comparison of industrial and capital market development in Brazil, Mexico, and the United States between 1830 and 1930. Using firm-level data, he finds that capital market development affected industrial composition and national economic performance. Specifically, Haber shows that
Ch. 12:
Finance and Growth: Theory and Evidence
909
when Brazil overthrew the monarchy in 1889 and formed the First Republic, it also dramatically liberalized restrictions on Brazilian financial markets. The liberalization gave more firms easier access to external finance. Industrial concentration fell and industrial production boomed. While Mexico also liberalized financial sector policies, the liberalization was much more mild under the Diaz dictatorship (1877–1911), which “. . . relied on the financial and political support of a small in-group of powerful financial capitalists” (p. 561). As a result, the decline in concentration and the increase in economic growth were much weaker in Mexico than it was in Brazil. Haber (1997) concludes that (1) international differences in financial development significantly impacted the rate of industrial expansion and (2) under-developed financial systems that restrict access to institutional sources of capital also impeded industrial expansion. In a recent firm-level study of China, Allen, Qian and Qian (2005) find that the linkages between the law, finance and growth are complex. Consistent with broad crosscountry findings discussed above, they find that poor legal protection of minority shareholder rights hinders the growth of publicly listed firms (as well as state-owned firms). However, private firms and firms owned by local governments have grown rapidly in absence of sound formal rules governing shareholder rights. This suggests the existence of effective alternative governance and financing mechanisms that promote firm growth. Additional evidence comes from Cull and Xu (2004), who find that private ownership is associated with firm reinvesting a greater proportion of their earnings than in firms with greater public sector ownership. Firm-level evidence from France also suggests the importance of well-functioning financial intermediaries for economic growth. Bertrand, Schoar and Thesmar (2004) examine the impact of deregulation in 1985 that eliminated government intervention in bank lending decisions and fostered greater competition in the credit market. They find that after deregulation, banks bailed out poorly performing firms less frequently, increased the cost of capital to poorly performing firms, and induced an increase in allocative efficiency across firms. This lowered industry concentration ratios and boosted both entry and exit rates for firms. While not directly tied to growth, the paper suggests that better functioning banks not only influence bank-firm relations they also exert a first-order impact on the structure and dynamics of product markets. In two classic studies, Cameron et al. (1967) and McKinnon (1973) study respectively (1) the historical relationships between banking development and the early stages of industrialization for England (1750–1844), Scotland (1750–1845), France (1800–1870), Belgium (1800–1875), Germany (1815–1870), Russia (1860–1914), and Japan (1868– 1914) and (2) the relationship between the financial system and economic development in Argentina, Brazil, Chile, Germany, Korea, Indonesia, and Taiwan in the post World War II period. This research does not use formal statistical analysis to resolve causality issues. Instead, the researchers carefully examine the evolution of the political, legal, policy, industrial, and financial systems of the country. The country-case studies document critical interactions among financial intermediaries, financial markets, government policies, and the financing of industrialization. While well-aware of the analytical limitations, these authors bring a wealth of country specific information to bear on the role
910
R. Levine
of finance in economic growth. Cameron (1967b) concludes that especially in Scotland and Japan, but also in Belgium, Germany, England, and Russia, the banking system played a positive, growth-inducing role.26 McKinnon (1973) interprets the mass of evidence emerging from his country-case studies as suggesting that better functioning financial systems support faster economic growth. Disagreement exists over many of these individual cases, and it is extremely difficult to isolate the importance of any single factor in the process of economic growth. Nonetheless, the body of country-studies suggests that, while the financial system responds to demands from the nonfinancial sector, well-functioning financial systems have, in some cases during some time periods, importantly spurred economic growth. 3.3. Industry and firm level studies of finance and growth To better understand the relationship between financial development and economic growth, researchers have employed both industry-level and firm-level data across a broad cross-section of countries. These studies seek to resolve causality issues and to document in greater detail the mechanisms, if any, through which finance influences economic growth. 3.3.1. Industry level analyses Consider first the influential study by Rajan and Zingales (1998, henceforth RZ). They argue that better-developed financial intermediaries and markets help overcome market frictions that drive a wedge between the price of external and internal finance. Lower costs of external finance facilitate firm growth and new firm formation. Therefore, industries that are naturally heavy users of external finance should benefit disproportionately more from greater financial development than industries that are not naturally heavy users of external finance. From this perspective, if researchers can identify which industries are “naturally heavy users” of external finance – i.e., if they can identify which industries rely heavily on external finance in an economy with few market frictions –
26 A valuable debate exists concerning the case of Scotland between 1750 and 1845 [Checkland (1975),
Cowen and Kroszner (1989)]. Scotland began the period with per capita income of less than one-half of England’s. By 1845, however, per capita income was about the same. While recognizing that the “. . . dominant political event affecting Scotland’s potentialities for economic development was the Union of 1707, which made Scotland an integral part of the United Kingdom” [Cameron (1967a, p. 60)], Cameron argues that Scotland’s superior banking system is one of the few noteworthy features that can help explain its comparatively rapid growth. Some researchers, however, suggest that England did not suffer from a dearth of financial services because nonfinancial enterprises provided financial services in England that Cameron’s (1967a) measures of formal financial intermediation omit. Others argue that Scotland had rich natural resources, a well-educated work force, access to British colonial markets, and started from a much lower level of income per capita than England. Consequently, it is not surprising that Scotland enjoyed a period of rapid convergence. Finally, still others disagree with the premise that Scotland had a well-functioning financial system and emphasize the deficiencies in the Scottish system [Sidney Pollard and Dieter Ziegler (1992)].
Ch. 12:
Finance and Growth: Theory and Evidence
911
then this establishes a natural test: Do industries that are naturally heavy users of external finance grow faster in economies with better developed financial systems? If they do, then this supports the view that financial development spurs growth by facilitating the flow of external finance. RZ assume that (1) financial markets in the U.S. are relatively frictionless, (2) in a frictionless financial system, technological factors influence the degree to which an industry uses external finance, and (3) the technological factors influencing external finance are constant (or reasonably constant) across countries. They then examine whether industries that are technologically more dependent on external finance – as defined by external use of funds in the U.S. – grow comparatively faster in countries that are more financially developed. This approach allows RZ (1) to study a particular mechanism, external finance, through which finance operates rather than simply assessing links between finance and growth and (2) to exploit within-country differences concerning industries. RZ develop a new methodology to examine the finance-growth relationship. Consider their formulation. αj Countryj + βl Industryl + γ Sharei,k Growthi,k = j
l
+ δ(Externalk · FDi ) + εi,k .
(4)
Growthi,k is the average annual growth rate of value added or the growth in the number of establishments, in industry k and country i, over the period 1980–90. Country and Industry are country and industry dummies, respectively. Sharei,k is the share of industry k in manufacturing in country i in 1980. Externalk is the fraction of capital expenditures not financed with internal funds for U.S. firms in the industry k between 1980–90. FDi is an indicator of financial development for country i. RZ interact the external dependence of an industry (External) with financial development (FD), where the estimated coefficient on the interaction, δ1 , is the focus of their analysis. Thus, if δ is significant and positive, then this implies that an increase in financial development (FDi ) will induce a bigger impact on industrial growth (Growthi,k ) if this industry relies heavily on external finance (Externalk ) than if this industry is not a naturally heavy user of external finance. They do not include financial development independently because they focus on within-country, within-industry growth rates. The dummy variables for industries and countries correct for country and industry specific characteristics that might determine industry growth patterns. RZ thus isolate the effect that the interaction of external dependence and financial development/structure has on industry growth rates relative to country and industry means. By including the initial share of an industry, this controls for a convergence effect: industries with a large share might grow more slowly, suggesting a negative sign on γ . RZ include the share in manufacturing rather than the level to focus on within-country, within-industry growth rates. RZ use data on 36 industries across 42 countries, though the U.S is dropped from the analyses since it is used to identify external dependence. To measure financial development, RZ examine (a) total capitalization, which equals the summation of stock market
912
R. Levine Table 6 Industry growth and financial development
Dependent variable: Growth of value added of industry k in country i, 1980–1990 Sharei,k of industry k in Externalk · Total Externalk · Accounting country i in 1980 capitalizationi standardsi −0.912 (0.246) −0.643 (0.204)
0.069 (0.023) 0.155 (0.034)
R2
Observations
0.29
1217
0.35
1067
Source: Rajan and Zingales (1998, Table 4). The table reports the results from the regression: Growthi,k =
j
αj Countryj +
βl Industryl + γ Sharei,k + δ1 (Externalk · FDi ) + εi,k .
l
Two regressions are reported corresponding to two values of FDi , Total capitalization and Account standards respectively. (Heteroscedasticity robust standard errors are reported in parentheses.) Externalk is the fraction of capital expenditures not financed with internal funds for U.S. firms in industry k between 1980–90. Total capitalization is stock market capitalization plus domestic credit. Accounting standards is an index of the quality of corporate financial reports.
capitalization and domestic credit as a share of GDP and (b) accounting standards. As RZ discuss, there are problems with these measures. Stock market capitalization does not capture the actual amount of capital raised in equity markets. Indeed, some countries provide tax incentives for firms to list, which artificially boosts stock market capitalization without indicating greater external financing or stock market development. Also, as discussed above, stock market capitalization does not necessarily reflect how well the market facilitates exchange. The accounting standards indicator is a rating of the quality of the annual financial reports issued by companies within a country. The highest value is 90. RZ use the accounting standards measure as a positive signal of the ease with which firms can raise external funds, while noting that it is not a direct measure of the actual amount of external funds that are raised. Beck and Levine (2002) confirm the RZ findings using alternative measures of financial development. As summarized in Table 6, RZ find that the coefficient estimate for the interaction between external dependence and total capitalization measure, Externalk · Total capitalizationi , is positive and statistically significant at the one-percent level. This implies that an increase in financial development disproportionately boosts the growth of industries that are naturally heavy users of external finance.27 27 Fisman and Love (2003b) critique the Rajan and Zingales (1998) methodology, arguing that it does not
accurately test whether financial development boosts growth in externally dependent industries. They argue
Ch. 12:
Finance and Growth: Theory and Evidence
913
RZ note that the economic magnitude is quite substantial. Compare Machinery, which is an industry at the 75th percentile of dependence (0.45), with Beverages, which has low dependence (0.08) and is at the 25th percentile of dependence. Now, consider Italy, which has high total capitalization (0.98) at the 75th percentile of the sample, and the Philippines, which is at the 25th percentile of total capitalization with a value of 0.46. Due to differences in financial development, the coefficient estimates predict that Machinery should grow 1.3 percent faster than Beverages in Italy in comparison to the Philippines.28 The actual difference is 3.4, so the estimated value of 1.3 is quite substantial. Thus, financial development has a substantial impact on industrial growth by influencing the availability of external finance. RZ conduct a large number of robustness checks and show that financial development influences industrial growth both through the expansion of existing establishments and through the formation of new establishments.29 Instead of examining the impact of banking sector development on the growth of externally dependent firms, recent work studies the impact of banking market structure and bank competition on industrial development. Cetorelli and Gambera (2001) examine the role played by banking sector concentration on firm access to capital. Using the RZ methodology, they show that bank concentration promotes the growth of industries that are naturally heavy users of external finance, but bank concentration has a depressing effect on overall economic growth. Claessens and Laeven (2005) disagree, however. They note that industrial organization theory indicates that market concentration is not necessarily a good proxy for the competitiveness of an industry. Consequently, they estimate an industrial organization-based measure of banking system competition. Claessens and Laeven (2005) then show that industries that are naturally heavy users of external finance grow faster in countries with more competitive banking systems. They find no
that the method simply tests whether financial intermediaries allow firms to respond to global shocks to growth opportunities, rather than the extent to which financial systems foster the growth of industries with an inherent financial dependence. 28 More specifically, let I indicate Italy, P indicate the Philippines, M indicate machinery, B indicate beverages, and g represent the growth of an industry in a country, then the differential growth rate of machinery and beverages in Italy from the difference in growth rate of machinery and beverages in the Philippines is as follows: [{g(I, M)} − {g(P, M)}] − [{g(I, B)} − {g(P, B)}]. Now, inserting estimates one obtains 1.3 = {0.069 · 0.45 · 0.98} − {0.069 · 0.45 · 0.46} − {0.069 · 0.08 · 0.98} − {0.069 · 0.08 · 0.46} . 29 Beck (2002, 2003) extends the work by RZ to examine the linkages between financial development and
international trade patterns. Beck (2002) develops a theoretical model in which higher levels of financial development provide countries with a comparative advantage in sectors with greater scale economies and presents econometric evidence consistent with this prediction. Using cross-industry and cross-country data on trade flows, Beck (2003) finds that countries with more developed financial systems tend to be net exporters in industries that are heavy users of external finance. The results of both papers are consistent with the view that financial development influences the structure of trade balances.
914
R. Levine
evidence that banking industry concentration explains industrial sector growth. The results support the view that banking sector competition fosters the provision of growth enhancing financial services. Building on RZ, Claessens and Laeven (2003) examine the joint impact of financial sector development and the quality of property rights protection on the access of firms to external finance and the allocation of resources. In particularly, they show that financial sector development hurts growth by hindering the access of firms to external finance and insecure property rights hurts growth by leading to a suboptimal allocation of resources by distorting firms into investing excessively in tangible assets. Thus, even when controlling for property rights protection, financial development continues to influence economic growth. This conclusion is different, however, from Johnson, McMillan and Woodruff’s (2002) study of post-communist countries. They find that property rights dominate access to external finance in explaining the degree to which firms reinvest their profits. Extending the RZ approach, Beck, Demirgüç-Kunt and Maksimovic (2004) highlight another channel linking finance and growth: removing impediments to small firms. They examine whether industries that are naturally composed of small firms grow faster in financially developed economies. More specifically, as in RZ, they assume that U.S. financial markets are relatively frictionless, so that the sizes of firms within industries in the U.S. reflect technological factors, not financial system frictions. Based on the U.S., they identify the benchmark average firm-size of each industry. Then, comparing across countries and industries, Beck, Demirgüç-Kunt and Maksimovic (2004) show that industries that are naturally composed of smaller firms grow faster in countries with better-developed financial systems. This result is robust to controlling for the RZ measure of external dependence. These results are consistent with the view that small firms face greater informational and contracting barriers to raising funds than large firms, so that financial development is particularly important for the growth of industries that, for technological reasons, are naturally composed of small firms. Using a different strategy, Wurgler (2000) also employs industry-level data to examine the relation between financial development and economic growth. Using industrylevel data across 65 countries for the period 1963–1995, he computes an investment elasticity that gauges the extent to which a country increases investment in growing industries and decreases investment in declining ones. This is an important contribution because it directly measures the degree to which each country’s financial system reallocates the flow of credit. Wurgler (2000) uses standard measures of financial development. He shows that countries with higher levels of financial development both increase investment more in growing industries and decrease investment more in declining industries than financial underdeveloped economies. 3.3.2. Firm level analyses of finance and growth Demirgüç-Kunt and Maksimovic (1998, henceforth DM) examine whether financial development influences the degree to which firms are constrained from investing in
Ch. 12:
Finance and Growth: Theory and Evidence
915
profitable growth opportunities. They focus on the use of long-term debt and external equity in funding firm growth. As in RZ, DM focuses on a particular mechanism through which finance influences growth: does greater financial development remove impediments to the exploitation of profitable growth opportunities. Rather than focusing on the external financing needs of an industry as in RZ, DM estimate the external financing needs of each individual firm in the sample. DM note that simple correlations between firms’ growth and financial development do not control for differences in the amount of external financing needed by firms in the same industry in different countries. These differences may arise because firms in different countries employ different technologies, because profit rates may differ across countries, or because investment opportunities and demand may differ. To control for these differences at the firm-level, DM calculate the rate at which each firm can grow using (1) only its internal funds and (2) only its internal funds and short-term borrowing. They then compute the percentage of firms that grow at rates that exceed each of these two estimated rates. This yields estimates of the proportion of firms in each economy relying on external financing to grow. The firm-level data consist of accounting data for the largest publicly traded manufacturing firms in 26 countries. Beck, Demirgüç-Kunt and Levine (2001) confirm the findings using an extended sample. DM estimate a firm’s potential growth rate using the textbook “percentage of sales” financial planning model [Higgins (1977)]. This approach relates a firm’s growth rate of sales to its need for investment funds, based on three simplifying assumptions. First, the ratio of assets used in production to sales is constant. Second, the firm’s profits per unit of sales are constant. Finally, the economic deprecation rate equals the accounting depreciation rate. Under these assumptions, the firm’s financing need in period t of a firm growing at gt percent per year is given by EFN t = gt · Assetst − (1 − gt ) · Earningst · bt
(5)
where EFN t is the external financing need and BT is the fraction of the firm’s earnings that are retained for reinvestment at time t. Earnings are calculated after interest and taxes. While the first term on the right-hand side of Equation (5) denotes the required investment for a firm growing at gt percent, the second term is the internally available funds for investment, taking the firms’ dividend payout as given. The short-term financed growth rate STFGt is the maximum growth rate that can be obtained if the firm reinvests all its earnings and obtains enough short-term external resources to maintain the ratio of its short-term liabilities to assets. To compute STFGt , we first replace total assets in (5) by assets that are not financed by new short-term credit, calculated as total assets times one minus the ratio of short-term liabilities to total assets. STFGt is then given by STFGt = ROLTCt /(1 − ROLTCt )
(6)
916
R. Levine
where ROLTCt is the ratio of earnings, after tax and interest, to long-term capital. The definition of STFG thus assumes that the firm does not access any long-term borrowings or sales of equity to finance its growth.30 DM then calculate the proportion of firms whose growth rates exceed the estimate of the maximum growth rate that can be financed by relying only on internal and shortterm financing, PROPORTION_FASTER. To analyze whether financial development spurs firm growth, DM run the following cross-country regression PROPORTION_FASTER = β1 FDi,t + β2 CV i,t + εi,t
(7)
where FD is financial development, CV is a set of control variables, and ε is the error term. To measure financial development, DM use (a) the ratio of market capitalization to GDP (Market Capitalization/GDP), (b) Turnover, which equals the total value of shares traded divided by market capitalization, and (c) Bank Assets/GDP, which equals the ratio of domestic assets of deposit banks divided by GDP. Thus, DM include all domestic assets of deposit banks, not just credit to the private sector. As control variables, DM experiment with different combinations of control variables, including economic growth, inflation, the average market to book value of firms in the economy, government subsidies to firms in the economy, the net fixed assets divided by total assets of firms in the economy, the level real per capita GDP, the law and order tradition of the economy. As summarized in Table 7, DM (1998) find that both banking system development and stock market liquidity are positively associated with the excess growth of firms. Thus, in countries with high Turnover and high Bank Assets/GDP a larger proportion of firms is growing at a level that requires access to external sources of long-term capital, holding other things constant. Note, consistent with LZ, the size of the domestic stock markets is not related to the excess growth of firms. After conducting a widearray of robustness checks, DM conclude that the proportion of firms that grow at rates exceeding the rate at which each firm can grow with only retained earnings and shortterm borrowing is positively associated with stock market liquidity and banking system size. Love (2003) and Beck, Demirgüç-Kunt and Maksimovic (2004) also use firm level data to examine whether financial development eases financing constraints, though they do not explicitly examine aggregate economic growth. Love finds that the sensitivity of investment to internal funds is greater in countries with more poorly developed financial 30 The estimates of internally financed growth (IFG) and short-term financed growth (STFG) are conserva-
tive. First, they assume that a firm utilizes the unconstrained sources of finance – trade credit in the case IFG and trade credit and short-term borrowing in the case of STFG – no more intensively than it is currently doing. Second, firms with spare capacities do not need to invest and may grow at a faster rate than predicted without accessing external resources. Third, the financial planning model abstracts from technical advances that reduce the requirements for investment capital. Thus, it may overstate the costs of growth and underestimate the maximum growth rate attainable using unconstrained sources of financing.
Ch. 12:
Finance and Growth: Theory and Evidence
917
Table 7 Excess growth of firms and external financing Dependent variable: Proportion of firms that grow faster than their predicted growth rate1 Market capitalization/GDP Turnover Bank assets/GDP Adj. R2 −0.106 (0.058)
0.311∗∗∗ (0.072)
0.162∗∗∗ (0.050)
0.48
Countries 26
Source: Demirguc-Kunt and Maksimovic (1998, Table V). (White’s heteroscedasticity consistent standard errors in parentheses.) ∗ ∗ ∗ Indicates statistical significance at the 1 percent level. Market capitalization/GDP: the value of domestic equities listed on domestic exchanges as a share of GDP. Turnover: the total value of trades of domestic shares on domestic exchanges as a share of market capitalization. Other regressors: rate of inflation, the law and order tradition of the economy, i.e., the extent to which citizens utilize existing legal system to mediate disputes and enforce contracts, growth rate of real GDP per capita, real GDP per capita, government subsidies to private industries and public enterprises as a share of GDP, and net fixes assets divided by total assets. Time period: the dependent variable is averaged over the 1986–1991 period. All regressors are averaged over the 1980–1985 period, data permitting. 1 The proportion of firms whose growth rates exceed the estimate of the maximum growth rate that can be financed by relying only on internal and short-term financing.
system. Greater financial development reduces the link between the availability of internal funds and investment. Thus, the paper is consistent with the findings of DM and RZ. The paper also shows that financial development is particularly effective at easing the constraints of small firms. Beck, Demirgüç-Kunt and Maksimovic (2004) use a different dataset and methodology to investigate the effect of financial development on easing the obstacles that firms face to growing faster. They show that financial development weakens the impact of various barriers to firm growth and that small firms benefit the most from financial development.31 In sum and consistent with the industry-level work by Beck, Demirgüç-Kunt and Maksimovic (2004), these firm-level studies indicate that financial development removes impediments to firm expansion and exerts a particularly beneficial impact on small firms. Dyck and Zingales (2004) provide additional firm-level evidence on the mechanisms through which financial development influences growth by examining whether financial development influences the private benefits of controlling a firm. If there are large private benefits of control, this implies that insiders can exploit their positions and help themselves at the expense of the firm. The resultant loss of corporate efficiency could have aggregate growth effects. Dyck and Zingales (2004) estimate the value of control in 393 control transactions across 39 countries over the period 1990–2000. They find
31 Kumar, Rajan and Zingales (2001) show that financial development is associated with larger firms, sug-
gesting that low levels of financial development constraint firm growth.
918
R. Levine
that the benefits of control are greater in countries with poorly-developed financial systems. While not linked with aggregate growth, this suggests that financial development improves the corporate governance of firms.32 3.4. Are bank- or market-based systems better? Evidence As noted earlier, Goldsmith (1969) asked whether (1) financial development influences economic growth and whether (2) financial structure – the mix of financial markets and intermediaries operating in an economy – affects economic growth. As we have seen, a growing body of evidence using very different methodologies and datasets find that financial development exerts a first-order impact on economic growth. We now turn to the empirical analysis of financial structure: Does having a bank-based or market-based financial system matter for economic growth? Much of the empirical work on financial structure over the last century involves studies of Germany and Japan as bank-based systems and the United States and the United Kingdom as market-based systems.33 As summarized by Allen and Gale (2000), Demirgüç-Kunt and Levine (2001a), and Stulz (2001), this research has produced illuminating insights into the functioning of these financial systems. Nonetheless, it is difficult to draw broad conclusions about the long-run growth effects of bank-based and market-based financial systems based on only four countries, especially four countries that have very similar long-run growth rates. Indeed, given the similarity of their longrun growth rates, many observers may conclude that differences in financial structure obviously did not matter much. Broadening the analysis to a wider array of national experiences is important for garnering greater information on the bank-based versus market-based debate. Recently, empirical research has expanded the study of financial structure to a much broader set of countries. Beck, Demirgüç-Kunt and Levine (2000, 2001) construct a large cross-country, time-series database on the mixture of financial markets and intermediaries across 150 countries for the period 1960–1995, data permitting. DemirgüçKunt and Levine (2001b) classify countries according to the degree to which they are bank-based or market-based. They also examine the evolution of financial structure across time and countries. They find that banks, nonbank financial intermediaries (insurance companies, pension funds, finance companies, mutual funds, etc.) and stock markets are larger, more active, and more efficient in richer countries and these components of the financial system grow as countries become richer over time. Also, as countries become richer, stock markets become more active and efficient relative to banks. There is a tendency, not without exceptions, for national financial systems to become more market-based as they become richer. Demirgüç-Kunt and Levine (2001b) 32 Dyck and Zingales (2002) stress the role of the media in influencing corporate managers. This work ex-
tends our conception of the institutions involved in exerting corporate control over firms. 33 See Goldsmith (1969), Hoshi, Kashyap and Sharfstein (1990), Allen and Gale (1995), Levine (1997),
Morck and Nakamura (1999), Weinstein and Yafeh (1998) and Wenger and Kaserer (1998).
Ch. 12:
Finance and Growth: Theory and Evidence
919
also show that countries with better functioning legal systems and institutions tend to have more market-based financial systems, a point also emphasized by Ergungor (2004). Turning to economic growth, an expanding body of empirical work uses these newly developed measures of financial structure and assortment of econometric methodologies to study the impact of financial structure and growth. This work employs the same methodologies used in the financial development and growth literature: (1) crosscountry regressions, including instrumental variables regressions, (2) industry-level studies, and (3) firm-level investigations. Since I have already reviewed these methodologies, this subsection succinctly discusses the findings on financial structure and growth. Using very different econometric methodologies, the literature finds, albeit with exceptions, astonishingly consistent results. First, in a cross-country context, there is no general rule that bank-based or marketbased financial systems are better at fostering growth. Levine (2002) finds that after controlling for the overall level of financial development, information on financial structure does not help in explaining cross-country differences in financial development. These results hold when using instrumental variables to control for simultaneity bias. This research also assesses whether bank-based systems are better at promoting growth in poor countries or countries with poor legal systems or otherwise weak institutions. Allowing for these possibilities, however, did not alter the conclusion: after controlling for overall financial development, cross-country comparisons do not suggest that distinguishing between bank-based and market-based financial systems is a first-order concern in understanding the process of economic growth. Tadesse (2002), however, argues that while market-based systems outperform bank-based systems among countries with developed financial sectors, bank-based systems are far better among countries with underdeveloped financial sectors. Second, using industry-level data, research finds that financially-dependent industries do not expand at higher rates in bank-based or market-based financial systems. Beck and Levine (2002) confirm that greater financial development accelerates the growth of financially dependent industries. Financial structure per se, however, does not help explain the differential growth rates of financially-dependent industries across countries. Third, firms’ access to external finance is not easier, and firms do not grow faster in either market-based or bank-based financial systems. Demirgüç-Kunt and Maksimovic (2002) extend their earlier study and show that overall financial development helps explain the excess growth of firms across countries, i.e., the proportion of firms that grow at rates exceeding the rate at which each firm can grow with only retained earnings and short-term borrowing is positively associated with overall financial development. However, the degree to which countries are bank-based or market-based does not help explain excess growth. I want to make two cautionary remarks about this research. First, these studies do not necessarily imply that institutional structure is unimportant for growth. Rather, the results may imply that there is not one optimal institutional structure for providing growth-enhancing financial functions to the economy [Merton and Bodie (2004)].
920
R. Levine
While the emergence of financial systems that ameliorate information, contracting, and transactions costs may be crucial for accelerating economic growth, the growthmaximizing mixture of markets and intermediaries may depend on legal, regulatory, political, and other factors that have not been adequately incorporated into current theoretical or empirical research. Second, recent research on financial structure and growth use aggregate, cross-country indicators of the degree to which countries are bank-based or market-based. These indicators may not sufficiently capture the comparative roles of banks and markets. They may not be sufficiently country-specific to gauge accurately national financial structure. Thus, the conclusion from these studies that financial structure is not a particularly useful indicator of the degree to which a financial system promotes growth must be viewed cautiously [Demirgüç-Kunt and Levine (2001a)]. Finally, Carlin and Mayer (2003) extend the recent work on financial structure and economic growth by examining the relationship between the structure of the financial system and types of activities conducted in different countries. They find a positive association between information disclosure (as measured by the effectiveness of the accounting system), the fragmentation of the banking system (as measured by low bank concentration), and the growth of equity-financed and skill-intensive industries. This is consistent with models by Allen and Gale (2000) and Boyd and Smith (1998) that emphasize that high technology firms require financial systems that allow for diverse views, such as equity markets rather than banks which provided more standardized monitoring. This result is also consistent with models by Dewatripont and Maskin (1995) that focus on renegotiations, where fragmented banking systems tend to impose short-term, tighter budget constraints. This may be more appropriate for new, higher-risk firms where the threat of bankruptcy must be credibly imposed. In contrast, concentrated banks with long-run relationships with firms can more easily renegotiate constructs and will have a correspondingly more difficult time credibly committing to not renegotiate. Thus, concentrated banks will tend to be associated with more mature, less risky firms. While not directly linked to aggregate economic growth, this sector-based work improves our understanding of the relationship between financial structure and the types of activities occurring in different economies. 3.5. Finance, income distribution, and poverty alleviation: evidence I conclude the review of empirical work on finance and growth by discussing some very recent research on whether financial development influences income distribution and poverty. As discussed above, theory offers conflicting predictions about the nature of the interactions between finance, income distribution, and poverty. In cross-country regressions, Beck, Demirgüç-Kunt and Levine (2004) examine whether the level of financial intermediary development influences (i) the growth rate of Gini coefficients of income inequality, (ii) the growth rate of the income of the poorest quintile of society, and (iii) the fraction of the population living in poverty. The results indicate that finance exerts a disproportionately large, positive impact on the poor and hence reduces income inequality. Even when controlling for the growth rate
Ch. 12:
Finance and Growth: Theory and Evidence
921
of real per capita GDP, the data indicate that (i) Gini coefficients fall more rapidly in countries with higher levels of financial intermediary development, (ii) the income of the poorest quintile grows faster than the national average with better-developed financial intermediaries, and (iii) the percentage of the population living on less than one or two dollars per day falls more rapidly in economies that have higher levels of financial development. These results hold when using instrumental variables to control for the endogenous determination of financial development and changes in income distribution and poverty alleviation. The findings lend cautious support to the view that financial development disproportionately boosts the income of the poor and reduces income inequality.34 At the same time, the extensive battery of methodological weaknesses associated with cross-country regression reviewed above can certainly be levied against these initial findings on finance, income distribution, and poverty. Consequently, applying diverse econometric techniques and datasets to bear on the question of whether financial development influences income distribution and poverty is likely to be an active area of research.
4. Conclusions This paper reviewed theoretical and empirical work on the relationship between financial development and economic growth. Theory illuminates many of the channels through which the emergence of financial instruments, markets and institutions affect – and are affected by – economic development. A growing body of empirical analyses, including firm-level studies, industry-level studies, individual country-studies, time-series studies, panel-investigations, and broad cross-country comparisons, demonstrate a strong positive link between the functioning of the financial system and longrun economic growth. While subject to ample qualifications and countervailing views noted throughout this article, the preponderance of evidence suggests that both financial intermediaries and markets matter for growth even when controlling for potential simultaneity bias. Furthermore, microeconomic-based evidence is consistent with the view that better developed financial systems ease external financing constraints facing firms, which illuminates one mechanism through which financial development influences economic growth. Theory and empirical evidence make it difficult to conclude that the financial system merely – and automatically – responds to economic activity, or that financial development is an inconsequential addendum to the process of economic growth. In the remainder of this section, I discuss broad areas needing additional research. In terms of theory, Section 2 raised several issues associated with modeling finance
34 See Clarke, Xu and Zou (2003), who study the cross-country relationship between financial intermediary
development and the level of the Gini coefficient, rather than the relationship between financial intermediary development the growth rate of the Gini coefficient.
922
R. Levine
and growth. Here I simply make one broad observation. Our understanding of finance and growth will be substantively advanced by the further modeling of the dynamic interactions between the evolution of the financial system and economic growth [Smith (2002)]. Existing work suggests that it is not just finance following industry. But, neither is there any reason to believe that it is just industry following finance. Thus, we need additional thought on the co-evolution of finance and growth. Technology innovation, for instance, may only foster growth in the presence of a financial system that can evolve effectively to help the economy exploit these new technologies. Furthermore, technological innovation itself may substantively affect the operation of financial systems by, for example, transforming the acquisition, processing, and dissemination of information. Moreover, the financial system may provide different services at different stages of economic development, so that the financial system needs to evolve if growth is to continue. These are mere conjectures and ruminations that I hope foster more careful thinking. In terms of empirical work, this paper continuously emphasized that all methods have their problems but that one problem plaguing the entire study of finance and growth pertains to the proxies for financial development. Theory suggests that financial systems influence growth by easing information and transactions costs and thereby improving the acquisition of information about firms, corporate governance, risk management, resource mobilization, and financial exchanges. Too frequently empirical measures of financial development do not directly measure these financial functions. While a growing number of country-specific studies develop financial development indicators more closely tied to theory, more work is needed on improving cross-country indicators of financial development. Much more research needs to be conducted on the determinants of financial development. To the extent that financial systems exert a first-order impact on economic growth, we need a fuller understanding of what determines financial development. There are at least two levels of analysis. There is a growing body of research that examines the direct laws, regulations, and macroeconomic policies shaping financial sector operations. There is a second research agenda that studies the political, cultural, and even geographic context shaping financial development. Some research examines how legal systems, regulations, and macroeconomic policies influence finance. LLSV (1997, 1998) show that laws and enforcement mechanisms that protect the rights of outside investors tend to foster financial development. Beck, Demirgüç-Kunt and Levine (2003b, 2005a) show that legal system adaptability is crucial. The financial needs of the economy are continuously changing, so that more flexible legal systems do a better job at promoting financial development than more rigid systems. Barth, Caprio and Levine (2001a, 2001b, 2004, 2005) and La Porta, Lopez-deSilanes and Shleifer (2005) show that regulations and supervisory practices that force accurate information disclosure and promote private sector monitoring, but do not grant regulators excessive power, boost the overall level of banking sector and stock market
Ch. 12:
Finance and Growth: Theory and Evidence
923
development.35 Monetary and fiscal policies may also affect the taxation of financial intermediaries and the provision of financial services [Bencivenga and Smith (1992), Huybens and Smith (1999), Roubini and Sala-i-Martin (1992, 1995)]. Indeed, Boyd, Levine and Smith (2001) show that inflation has a large – albeit non-linear – impact on both stock market and bank development. At a more primitive level, some research studies the forces shaping the laws, regulations, and institutions underlying financial development. LLSV (1998) stress that historically-determined differences in legal tradition shape the laws governing financial transactions. Haber (2004), Haber, Maurer and Razo (2003), Pagano and Volpin (2001), Roe (1994), and Rajan and Zingales (2003) focus on how political economy forces shape national policies toward financial development. Guiso, Sapienza and Zingales (2004) examine the role of social capital in shaping financial systems, while Stulz and Williamson (2003) stress the role of religion in influencing national approaches to financial development. Finally, some scholars emphasize the impact of geographical endowments on the formation of long-lasting institutions that form the foundations of financial systems [Engerman and Sokoloff (1997, 2002), Acemoglu, Johnson and Robinson (2001), Beck, Demirgüç-Kunt and Levine (2003a), Easterly and Levine (2003)]. This broad spectrum of work suggests that political, legal, cultural, and even geographical factors influence the financial system and that much more work is required to better understand the role of financial factors in the process of economic growth.
Acknowledgements Philippe Aghion, Thorsten Beck, John Boyd, Maria Carkovic, Asli Demirguc-Kunt, Steven Durlauf, John Kareken, Luc Laeven, Raghu Rajan, Bruce Smith, and Luigi Zingales provided helpful comments.
References Acemoglu, D., Aghion, P., Zilibotti, F. (2003). “Distance to frontier, selection, and economic growth”. Working Paper No. 9066. National Bureau of Economic Growth. Acemoglu, D., Johnson, S., Robinson, J.A. (2001). “The colonial origins of comparative development: An empirical investigation”. American Economic Review 91, 1369–1401.
35 Beck, Demirgüç-Kunt and Levine (2003c, 2005b) go on to show that bank supervisory practices that
force accurate information disclosure ease external financing constraints facing firms, while countries that grant substantial power to government controlled regulators actually make external financing constraints more severe by increasing the degree of corruption in bank lending. Caprio, Laeven and Levine (2003) show that legal protection of shareholders is more effective at boosting the valuation of banks than strong official bank regulation and supervision. Bodenhorn (2003) examines the influences of political forces, fiscal demands, and regulations on the development of banking sectors in individual states of the United States.
924
R. Levine
Acemoglu, D., Zilibotti, F. (1997). “Was Prometheus unbound by chance? Risk, diversification, and growth”. Journal of Political Economy 105, 709–775. Aghion, P., Bolton, P. (1997). “A trickle-down theory of growth and development with debt overhang”. Review of Economic Studies 64, 151–172. Aghion, P., Dewatripont, M., Rey, P. (1999). “Competition, financial discipline and growth”. Review of Economic Studies 66, 825–852. Aghion, P., Howitt, P. (1998). Endogenous Economic Growth Theory. MIT Press, Cambridge, MA. Aghion, P., Howitt, P., Mayer-Foulkes, D. (2005). “The effect of financial development on convergence: Theory and evidence”. Quarterly Journal of Economics, 323–351. Aghion, P., Angeletos, M., Banerjee, A., Manova, K. (2004). “Volatility and growth: The role of financial development”. Mimeo. Department of Economics, Harvard University. Allen, F. (1990). “The market for information and the origin of financial intermediaries”. Journal of Financial Intermediation 1, 3–30. Allen, F., Gale, D. (1995). “A welfare comparison of the German and U.S. financial systems”. European Economic Review 39, 179–209. Allen, F., Gale, D. (1997). “Financial markets, intermediaries, and intertemporal smoothing”. Journal of Political Economy 105, 523–546. Allen, F., Gale, D. (1999). “Diversity of opinion and financing of new technologies”. Journal of Financial Intermediation 8, 68–89. Allen, F., Gale, D. (2000). Comparing Financial Systems. MIT Press, Cambridge, MA. Allen, F., Qian, J., Qian, M. (2005). “Law, finance, and economic growth in China”. Journal of Financial Economics. In press. Arellano, M., Bond, S. (1991). “Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations”. Review of Economic Studies 58, 277–297. Arellano, M., Bover, O. (1995). “Another look at the instrumental-variable estimation of error-components models”. Journal of Econometrics 68, 29–52. Arestis, P., Demetriades, P. (1997). “Financial development and economic growth: Assessing the evidence”. Economic Journal 107, 783–799. Arestis, P., Demetriades, P.O., Luintel, K.B. (2001). “Financial development and economic growth: The role of stock markets”. Journal of Money, Credit, and Banking 33, 16–41. Atje, R., Jovanovic, B. (1993). “Stock markets and development”. European Economic Review 37, 632–640. Bagehot, W. (1873). Lombard Street, 1962 ed. Irwin, Homewood, IL. Banerjee, A., Newman, A. (1993). “Occupational choice and the process of development”. Journal of Political Economy 101, 274–298. Barth, J.R., Caprio, G. Jr., Levine, R. (2001a). “The regulation and supervision of banks around the world: A new database”. In: Litan, R.E., Herring, R. (Eds.), Brooking–Wharton Papers on Financial Services. Brookings Institution, Washington, DC, pp. 83–250. Barth, J.R., Caprio, G. Jr., Levine, R. (2001b). “Banking systems around the globe: Do regulations and ownership affect performance and stability?”. In: Mishkin, F.S. (Ed.), Prudential Supervision: What Works and What Doesn’t. University of Chicago Press, Chicago, IL, pp. 31–96. Barth, J.R., Caprio, G. Jr., Levine, R. (2004). “Bank regulation and supervision: What works best?”. Journal of Financial Intermediation 13, 205–248. Barth, J.R., Caprio, G. Jr., Levine, R. (2005). Rethinking Bank Supervision and Regulation: Until Angels Govern. Cambridge University Press, Cambridge, UK. Beck, T. (2002). “Financial development and international trade: Is there a link?”. Journal of International Economics 57, 107–131. Beck, T. (2003). “Financial dependence and international trade”. Review of International Economics 11, 296– 316. Beck, T., Demirgüç-Kunt, A., Levine, R. (2000). “A new database on financial development and structure”. World Bank Economic Review 14, 597–605.
Ch. 12:
Finance and Growth: Theory and Evidence
925
Beck, T., Demirgüç-Kunt, A., Levine, R. (2001). “The financial structure database”. In: Demirgüç-Kunt, A., Levine, R. (Eds.), Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development. MIT Press, Cambridge, MA, pp. 17–80. Beck, T., Demirgüç-Kunt, A., Levine, R. (2003a). “Law, endowments, and finance”. Journal of Financial Economics 70, 137–181. Beck, T., Demirgüç-Kunt, A., Levine, R. (2003b). “Law and finance: Why does legal origin matter?””. Journal of Comparative Economics 31, 653–675. Beck, T., Demirgüç-Kunt, A., Levine, R. (2003c). “Bank supervision and corporate finance”. Working Paper No. 9620. National Bureau of Economic Research. Beck, T., Demirgüç-Kunt, A., Levine, R. (2003d). “Bank concentration and crises”. Working Paper No. 9921. National Bureau of Economic Research. Beck, T., Demirgüç-Kunt, A., Levine, R. (2004). “Finance, inequality and poverty: Cross-country evidence”. Mimeo. University of Minnesota (Carlson School of Management). Beck, T., Demirgüç-Kunt, A., Levine, R. (2005a). “Law and firms’ access to finance”. American Law and Economics Review 7, 211–252. Beck, T., Demirgüç-Kunt, A., Levine, R. (2005b). “Bank supervision and corruption in lending”. Working Paper No. 11498. National Bureau of Economic Research. Beck, T., Demirgüç-Kunt, A., Maksimovic, V. (2004). “Financial and legal constraints to firm growth: Does size matter?”. Journal of Finance 60, 137–177. Beck, T., Levine, R. (2002). “Industry growth and capital allocation: Does having a market- or bank-based system matter?”. Journal of Financial Economics 64, 147–180. Beck, T., Levine, R. (2004). “Stock markets, banks and growth: Panel evidence”. Journal of Banking and Finance, 423–442. Beck, T., Levine, R., Loayza, N. (2000). “Finance and the sources of growth”. Journal of Financial Economics 58, 261–300. Bekaert, G., Harvey, C.R., Lundblad, C. (2001). “Emerging equity markets and economic development”. Journal of Development Economics 66, 465–504. Bekaert, G., Harvey, C.R., Lundblad, C. (2005). “Does financial liberalization spur growth?”. Journal of Financial Economics. In press. Bencivenga, V.R., Smith, B.D. (1991). “Financial intermediation and endogenous growth”. Review of Economics Studies 58, 195–209. Bencivenga, V.R., Smith, B.D. (1992). “Deficits, inflation and the banking system in developing countries: The optimal degree of financial repression”. Oxford Economic Papers 44, 767–790. Bencivenga, V.R., Smith, B.D. (1993). “Some consequences of credit rationing in an endogenous growth model”. Journal of Economic Dynamics and Control 17, 97–122. Bencivenga, V.R., Smith, B.D., Starr, R.M. (1995). “Transactions costs, technological choice, and endogenous growth”. Journal of Economic Theory 67, 53–177. Benhabib, J., Spiegel, M.M. (2000). “The role of financial development in growth and investment”. Journal of Economic Growth 5, 341–360. Berle, A.A., Means, G.C. (1932). The Modern Corporation and Private Property. Harcourt Brace Jovanovich, New York. Bhattacharya, S., Pfleiderer, P. (1985). “Delegated portfolio management”. Journal of Economic Theory 36, 1–25. Berger, A.N., Hasan, I., Klapper, L.F. (2005). “Further evidence on the link between finance and growth: An international analysis of community banking and economic performance”. Journal of Financial Services Research 25 (2–3), 169–202. Bertrand, M., Schoar, A.S., Thesmar, D. (2004). “Banking deregulation and industry structure: Evidence from the French banking reforms of 1985”. Discussion Paper No. 4488. Centre for Economic Policy Research. Bhide, A. (1993). “The hidden costs of stock market liquidity”. Journal of Financial Economics 34, 1–51. Black, S.W., Moersch, M. (1998). “Financial structure, investment and economic growth in OECD countries”. In: Black, S.W., Moersch, M. (Eds.), Competition and Convergence in Financial Markets: The German and Anglo-American Models. North-Holland, New York, pp. 157–174.
926
R. Levine
Blackburn, K., Hung, V.T.Y. (1998). “A theory of growth, financial development, and trade”. Economica 65, 107–124. Bodenhorn, H. (2003). State Banking in Early America: A New Economic History. Cambridge University Press, New York. Boot, A.W.A., Greenbaum, S.J., Thakor, A. (1993). “Reputation and discretion in financial contracting”. American Economic Review 83, 1165–1183. Boot, A.W.A., Thakor, A. (1997). “Financial system architecture”. Review of Financial Studies 10, 693–733. Boot, A.W.A., Thakor, A. (2000). “Can relationship banking survive competition?”. Journal of Finance 55, 679–713. Boyd, J.H., Levine, R., Smith, B.D. (2001). “The impact of inflation on financial sector performance”. Journal of Monetary Economics 47, 221–248. Boyd, J.H., Prescott, E.C. (1986). “Financial intermediary-coalitions”. Journal of Economics Theory 38, 211– 232. Boyd, J.H., Smith, B.D. (1992). “Intermediation and the equilibrium allocation of investment capital: Implications for economic development”. Journal of Monetary Economics 30, 409–432. Boyd, J.H., Smith, B.D. (1994). “How good are standard debt contracts? Stochastic versus nonstochastic monitoring in a costly state verification environment”. Journal of Business 67, 539–562. Boyd, J.H., Smith, B.D. (1996). “The co-evolution of the real and financial sectors in the growth process”. World Bank Economic Review 10, 371–396. Boyd, J.H., Smith, B.D. (1998). “The evolution of debt and equity markets in economic development”. Economic Theory 12, 519–560. Calomiris, C., Kahn, C. (1991). “The role of demandable debt in structuring optimal banking arrangements”. American Economic Review 81, 497–513. Cameron, R. (1967a). “Scotland, 1750–1845”. In: Cameron, R., et al. (Eds.), Banking in the Early Stages of Industrialization: A Study in Comparative Economic History. Oxford University Press, New York, pp. 60–99. Cameron, R. (1967b). “Conclusion”. In: Cameron, R., et al. (Eds.), Banking in the Early Stages of Industrialization: A Study in Comparative Economic History. Oxford University Press, New York, pp. 290–321. Cameron, R., Crisp, O., Patrick, H.T., Tilly, R. (1967). Banking in the Early Stages of Industrialization: A Study in Comparative Economic History. Oxford University Press, New York. Caprio, G. Jr., Laeven, L., Levine, R. (2003). “Governance and bank valuation”. National Bureau of Economic Research Working Paper No. 10158. Caprio, G. Jr., Levine, R. (2002). “Corporate governance in finance: Concepts and international observations”. In: Litan, R.E., Pomerleano, M., Sundararajan, V. (Eds.), Financial Sector Governance: The Roles of the Public and Private Sectors. The Brookings Institution, Washington, DC, pp. 17–50. Carlin, W., Mayer, C. (2003). “Finance, investment, and growth”. Journal of Financial Economics 69, 191– 226. Carosso, V. (1970). Investment Banking in America. Harvard University Press, Cambridge, MA. Cetorelli, N., Gambera, M. (2001). “Banking structure, financial dependence and growth: International evidence from industry data”. Journal of Finance 56, 617–648. Charkham, J. (1994). Keeping Good Company: A Study of Corporate Governance in Five Countries. Clarendon Press, Oxford. Chakraborty, S., Ray, R. (2004). “Bank-based versus market-based financial systems: A growth-theoretic analysis”. Mimeo. University of Oregon (Department of Economics). Checkland, S.G. (1975). Scottish Banking: A History, 1695–1973. Collins, Glasgow. Christopoulos, D.K., Tsionas, E.G. (2004). “Financial development and economic growth: Evidence from panel unit root and cointegration tests”. Journal of Development Economics 73, 55–74. Claessens, S., Laeven, L. (2003). “Financial development, property rights, and growth”. Journal of Finance 58, 2401–2436. Claessens, S., Laeven, L. (2005). “Financial dependence, banking sector competition, and economic growth”. Journal of the European Economic Association 3, 179–207.
Ch. 12:
Finance and Growth: Theory and Evidence
927
Claessens, S., Djankov, S., Fan, J., Lang, L. (2002). “Disentangling the incentive and entrenchment effects of large shareholdings”. Journal of Finance 57, 2741–2771. Clarke, G., Xu, L.C., Zou, H. (2003). “Finance and income inequality, test of alternative theories”. World Bank Policy Research Working Paper, #2984. Coase, R. (1937). “The nature of the firm”. Economica 4, 386–405. Cowen, T., Kroszner, R. (1989). “Scottish banking before 1845: A model for laissez-faire?”. Journal of Money, Credit, and Banking 21, 221–231. Cull, R., Xu, L. (2004). “Institutions, ownership, and finance: The determinants of profit reinvestment among Chinese firms”. World Bank mimeo. DeAngelo, H., Rice, E. (1983). “Anti-takeover amendments and stockholder wealth”. Journal of Financial Economics 11, 329–360. De Gregorio, J. (1996). “Borrowing constraints, human capital accumulation, and growth”. Journal of Monetary Economics 37, 49–71. Dehejia, R., Lleras-Muney, A. (2003). “Why does financial development matter? The United States from 1900 to 1940”. Working Paper No. 9551. National Bureau of Economic Research. De la Fuente, A., Marin, J.M. (1996). “Innovation, bank monitoring, and endogenous financial development”. Journal of Monetary Economics 38, 269–301. DeLong, J.B. (1991). “Did Morgan’s men add value? An economist’s perspective on finance capitalism”. In: Temin, P. (Ed.), Inside the Business Enterprise: Historical Perspectives on the Use of Information. University of Chicago Press, Chicago, pp. 205–236. Demetriades, P., Hussein, K. (1996). “Does financial development cause economic growth? Time series evidence from 16 countries”. Journal of Development Economics 51, 387–411. Demirgüç-Kunt, A., Laeven, L., Levine, R. (2004). “Regulations, market structure, institutions, and the cost of financial intermediation”. Journal of Money, Credit, and Banking 36, 593–622. Demirgüç-Kunt, A., Levine, R. (1996). “Stock market development and financial intermediaries: Stylized facts”. World Bank Economic Review 10, 291–322. Demirgüç-Kunt, A., Levine, R. (2001a). “Financial structure and economic growth: Perspectives and lessons”. In: Demirgüç-Kunt, A., Levine, R. (Eds.), Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development. MIT Press, Cambridge, MA, pp. 3–14. Demirgüç-Kunt, A., Levine, R. (2001b). “Bank-based and market-based financial systems: Cross-country comparisons”. In: Demirgüç-Kunt, A., Levine, R. (Eds.), Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development. MIT Press, Cambridge, MA, pp. 81– 140. Demirgüç-Kunt, A., Levine, R. (2001c). Financial Structures and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development. MIT Press, Cambridge, MA. Demirgüç-Kunt, A., Maksimovic, V. (1996). “Stock market development and firm financing choices”. World Bank Economic Review 10, 341–370. Demirgüç-Kunt, A., Maksimovic, V. (1998). “Law, finance, and firm growth”. Journal of Finance 53, 2107– 2137. Demirgüç-Kunt, A., Maksimovic, V. (2001). “Firms as financial intermediaries: Evidence from trade credit data”. World Bank mimeo. Demirgüç-Kunt, A., Maksimovic, V. (2002). “Funding growth in bank-based and market-based financial systems: Evidence from firm level data”. Journal of Financial Economic 65, 337–363. Devereux, M.B., Smith, G.W. (1994). “International risk sharing and economic growth”. International Economic Review 35, 535–550. Dewatripont, M., Maskin, E. (1995). “Credit efficiency in centralized and decentralized economies”. Review of Economic Studies 62, 541–555. Diamond, D.W. (1984). “Financial intermediation and delegated monitoring”. Review of Economic Studies 51, 393–414. Diamond, D.W. (1991). “Monitoring and reputation: The choice between bank loans and directly placed debt”. Journal of Political Economy 99, 689–721.
928
R. Levine
Diamond, D.W., Dybvig, P.H. (1983). “Bank runs, deposit insurance, and liquidity”. Journal of Political Economy 91, 401–419. Diamond, D.W., Rajan, R. (2001). “Liquidity risk, liquidity creation, and financial fragility: A theory of banking”. Journal of Political Economy 109, 289–327. Diamond, D.W., Verrecchia, R.E. (1982). “Optimal managerial contracts and equilibrium security prices”. Journal of Finance 37, 275–287. Dyck, A., Zingales, L. (2002). “The corporate governance role of the media”. In: Islam, R. (Ed.), The Right to Tell: The Role of Mass Media in Economic Development. World Bank, Washington, DC, pp. 107–140. Dyck, A., Zingales, L. (2004). “Private benefits of control: An international comparison”. Journal of Finance 59, 537–600. Easterly, W., Levine, R. (2001). “It’s not factor accumulation: Stylized facts and growth”. World Bank Economic Review 15, 177–219. Easterly, W., Levine, R. (2003). “Tropics, germs, and crops: How endowments influence economic development”. Journal of Monetary Economics 50, 3–39. Edison, H., Levine, R., Ricci, L., Slok, T. (2002). “International financial integration and economic growth”. Journal of International Money and Finance 21, 749–776. Engerman, S.L., Sokoloff, K.L. (1997). “Factor endowments, institutions, and differential paths of growth among New World economies: A view from economic historians of the United States”. In: Haber, S. (Ed.), How Latin America Fell Behind. Stanford University Press, Stanford, CA, pp. 260–304. Engerman, S.L., Sokoloff, K.L. (2002). “Factor endowments, inequality, and paths of development among New World economies”. National Bureau of Economic Research Working Paper No. 9259. Ergungor, O.E. (2004). “Market- vs. bank-based financial systems: Do rights and regulations really matter?”. Journal of Banking and Finance 28, 2869–2887. Fama, E., Jensen, M. (1983a). “Separation of ownership and control”. Journal of Law and Economics 27, 301–325. Fama, E., Jensen, M. (1983b). “Agency problems and residual claims”. Journal of Law and Economics 26, 327–349. Fink, G., Haiss, P., Hristoforova, S. (2003). “Bond markets and economic growth”. Working Paper 49. Research Institute for European Affairs. Fisman, R.J., Love, I. (2003a). “Trade credit, financial intermediary development, and industry growth”. Journal of Finance 58, 353–374. Fisman, R.J., Love, I. (2003b). “Financial development and growth revisited”. National Bureau of Economic Research Paper No. 9582. Flannery, M. (1994). “Debt maturity and the deadweight cost of leverage: Optimally financing banking firms”. American Economic Review 84, 320–331. Furfine, C.H. (2001). “Banks as monitors of other banks: Evidence from the overnight federal funds market”. Journal of Business 74, 33–57. Gale, D., Hellwig, M. (1985). “Incentive-compatible debt contracts: The one-period problem”. Review of Economics Studies 52, 647–663. Galetovic, A. (1996). “Specialization, intermediation and growth”. Journal of Monetary Economics 38, 549– 559. Galor, O., Zeira, J. (1993). “Income distribution and macroeconomics”. Review of Economic Studies 60, 35–52. Gerschenkron, A. (1962). Economic Backwardness in Historical Perspective – A Book of Essays. Harvard University Press, Cambridge, MA. Goldsmith, R.W. (1969). Financial Structure and Development. Yale University Press, New Haven, CT. Gorton, G., Pennacchi, G. (1990). “Financial intermediaries and liquidity creation”. Journal of Finance 45, 49–71. Greenwood, J., Jovanovic, B. (1990). “Financial development, growth, and the distribution of income”. Journal of Political Economy 98, 1076–1107. Greenwood, J., Smith, B. (1996). “Financial markets in development, and the development of financial markets”. Journal of Economic Dynamics and Control 21, 145–181.
Ch. 12:
Finance and Growth: Theory and Evidence
929
Grossman, S.J., Hart, O. (1980). “Takeover bids, the free-rider problem, and the theory of the corporation”. Bell Journal of Economics 11, 42–64. Grossman, S.J., Hart, O. (1986). “The cost and benefit of ownership: A theory of lateral and vertical integration”. Journal of Political Economy 94, 691–719. Grossman, S.J., Miller, M.H. (1988). “Liquidity and market structure”. Journal of Finance 43, 617–633. Grossman, S.J., Stiglitz, J. (1980). “On the impossibility of informationally efficient markets”. American Economic Review 70, 393–408. Guiso, L., Sapienza, P., Zingales, L. (2002). “Does local financial development matter?”. National Bureau of Economic Research Working Paper No. 8922. Guiso, L., Sapienza, P., Zingales, L. (2004). “The role of social capital in financial development”. American Economic Review 94, 526–556. Gurley, J.G., Shaw, E.S. (1955). “Financial aspects of economic development”. American Economic Review 45, 515–538. Haber, S.H. (1991). “Industrial concentration and the capital markets: A comparative study of Brazil, Mexico, and the United States, 1830–1930”. Journal of Economic History 51, 559–580. Haber, S.H. (1997). “Financial markets and industrial development: A comparative study of governmental regulation, financial innovation and industrial structure in Brazil and Mexico, 1840–1940”. In: Haber, S. (Ed.), How Latin America Fell Behind? Stanford University Press, Stanford, CA, pp. 146–178. Haber, S.H. (2004). “Political competition and economic growth: Lessons from the political economy of bank regulation in the United States and Mexico”. Mimeo. Stanford University. Haber, S.H. (2005). “Mexico’s experiment with bank privatization and liberalization, 1991–2004”. Journal of Banking and Finance. In press. Haber, S.H., Maurer, N., Razo, A. (2003). The Politics of Property Rights: Political Instability, Credible Commitments, and Economic Growth in Mexico. Cambridge University Press. Harrison, P., Sussman, O., Zeira, J. (1999). “Finance and growth: Theory and evidence”. Mimeo. Federal Reserve Board (Division of Research and Statistics), Washington, DC. Hellwig, M. (1991). “Banking, financial intermediation, and corporate finance”. In: Giovanni, A., Mayer, C. (Eds.), European Financial Integration. Cambridge University Press, Cambridge, UK, pp. 35–63. Henry, P. (2000). “Do stock market liberalizations cause investment booms?”. Journal of Financial Economics 58, 301–334. Henry, P. (2003). “Capital account liberalization, the cost of capital, and economic growth”. American Economic Review 93, 91–96. Hicks, J. (1969). A Theory of Economic History. Clarendon Press, Oxford. Higgins, R.C. (1977). “How much growth can a firm afford?”. Financial Management 6, 3–16. Holmstrom, B., Tirole, J. (1993). “Market liquidity and performance monitoring”. Journal of Political Economy 101, 678–709. Holmstrom, B., Tirole, J. (1998). “Private and public supply of liquidity”. Journal of Political Economy 106, 1–40. Hoshi, T., Kashyap, A., Sharfstein, D. (1990). “Bank monitoring and investment: Evidence from the changing structure of Japanese corporate banking relationships”. In: Hubbard, R.G. (Ed.), Asymmetric Information, Corporate Finance and Investment. University of Chicago Press, Chicago, pp. 105–126. Hoshi, T., Kashyap, A., Sharfstein, D. (1991). “Corporate structure, liquidity, and investment: Evidence from Japanese panel data”. Quarterly Journal of Economics 27, 33–60. Huybens, E., Smith, R. (1999). “Inflation, financial markets, and long-run real activity”. Journal of Monetary Economics 43, 283–315. Jacklin, C. (1987). “Demand deposits, trading restrictions, and risk sharing”. In: Prescott, E.D., Wallace, N. (Eds.), Contractual Arrangements for Intertemporal Trade. University of Minnesota Press, Minneapolis, pp. 26–47. Jacoby, H.G. (1994). “Borrowing constraints and progress through school: Evidence from Peru”. Review of Economics and Statistics 76, 151–160. Jayaratne, J., Strahan, P.E. (1996). “The finance-growth nexus: Evidence from bank branch deregulation”. Quarterly Journal of Economics 111, 639–670.
930
R. Levine
Jayaratne, J., Strahan, P.E. (1998). “Entry restrictions, industry evolution, and dynamic efficiency: Evidence from commercial banking”. Journal of Law and Economics 41, 239–273. Jensen, M. (1993). “The modern industrial revolution, exit, and the failure of internal control systems”. Journal of Finance 48, 831–880. Jensen, M., Meckling, W.R. (1976). “Theory of the firm, managerial behavior, agency costs and ownership structure”. Journal of Financial Economics 3, 305–360. Jensen, M., Murphy, K. (1990). “Performance pay and top management incentives”. Journal of Political Economy 98, 225–263. Johnson, S., McMillan, J., Woodruff, C. (2002). “Property rights and finance”. American Economic Review 92, 1335–1356. Jorgenson, D.W. (1995). Productivity. MIT Press, Cambridge, MA. Jorgenson, D.W. (2005). “Accounting for growth in the information age”. In: Aghion, P., Durlauf, S. (Eds.), Handbook of Economic Growth, vol. 1A. North-Holland/Elsevier, Amsterdam. This volume, Chapter 10. Jung, W.S. (1986). “Financial development and economic growth: International evidence”. Economic Development and Cultural Change 34, 333–346. Kashyap, A., Stein, J., Rajan, R. (2002). “Banks as providers of liquidity: An explanation for the co-existence of lending and deposit-taking”. Journal of Finance 57, 33–73. King, R.G., Levine, R. (1993a). “Finance and growth: Schumpeter might be right”. Quarterly Journal of Economics 108, 717–738. King, R.G., Levine, R. (1993b). “Finance, entrepreneurship, and growth: Theory and evidence”. Journal of Monetary Economics 32, 513–542. King, R.G., Levine, R. (1993c). “Financial intermediation and economic development”. In: Mayer, C., Vives, X. (Eds.), Financial Intermediation in the Construction of Europe. Centre for Economic Policy Research, London, pp. 156–189. King, R.G., Levine, R. (1994). “Capital fundamentalism, economic development and economic growth”. Carnegie–Rochester Conference Series on Public Policy 40, 259–292. King, R.G., Plosser, C.I. (1986). “Money as the mechanism of exchange”. Journal of Monetary Economics 17, 93–115. Klein, M., Olivei, G. (2001). “Capital account liberalization, financial depth, and economic growth”. Unpublished. Tufts University, Somerville, MA. Kumar, K.B., Rajan, R.G., Zingales, L. (2001). “What determines firm size”. Mimeo. University of Chicago. Krebs, T. (2003). “Human capital risk and economic growth”. Quarterly Journal of Economics 118, 709–744. Kyle, A.S. (1984). “Market structure, information, futures markets, and price formation”. In: Storey, G.G., Schmitz, A., Sarris, A.H. (Eds.), International Agricultural Trade: Advanced Readings in Price Formation, Market Structure, and Price Instability. Westview, Boulder, CO. Laeven, L. (2001). “Insider lending: The case of Russia”. Journal of Comparative Economics 29, 207–229. Laeven, L., Levine, R. (2005). “Is there a diversification discount in financial conglomerates?”. Journal of Financial Economics. In press. Lamoreaux, N. (1994). Insider Lending: Banks, Personal Connections, and Economic Development in Industrial New England. Cambridge University Press, New York. La Porta, R., Lopez-de-Silanes, F., Shleifer, A. (2002). “Government ownership of commercial banks”. Journal of Finance 57, 265–301. La Porta, R., Lopez-de-Silanes, F., Shleifer, A. (2005). “What works in securities laws?”. Journal of Finance. In press. La Porta, R., Lopez-de-Silanes, F., Zamarripa, G. (2003). “Related lending”. Quarterly Journal of Economics. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W. (1997). “Legal determinants of external finance”. Journal of Finance 52, 1131–1150. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W. (1998). “Law and finance”. Journal of Political Economy 106, 1113–1155. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R. (1999). “Corporate ownership around the world”. Journal of Finance 54, 471–517.
Ch. 12:
Finance and Growth: Theory and Evidence
931
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R. (2000). “Investor protection and corporate governance”. Journal of Financial Economics 58, 3–27. Levhari, D., Srinivasan, T.N. (1969). “Optimal savings under uncertainty”. Review of Economic Studies 35, 153–163. Levine, R. (1991). “Stock markets, growth, and tax policy”. Journal of Finance 46, 1445–1465. Levine, R. (1997). “Financial development and economic growth: Views and agenda”. Journal of Economic Literature 35, 688–726. Levine, R. (1998). “The legal environment, banks, and long-run economic growth”. Journal of Money, Credit, and Banking 30, 596–613. Levine, R. (1999). “Law, finance, and economic growth”. Journal of Financial Intermediation 8, 36–67. Levine, R. (2002). “Bank-based or market-based financial systems: Which is better?”. Journal of Financial Intermediation 11, 398–428. Levine, R. (2003). “Napoleon, bourses, and growth: With a focus on Latin America”. In: Azfar, O., Cadwell, C. (Eds.), Market Augmenting Government. University of Michigan Press, Ann Arbor, MI, pp. 49–85. Levine, R. (2004). “Denying foreign bank entry: Implications for bank interest margins”. In: Ahumada, L.A., Fuentes, J.R. (Eds.), Bank Market Structure and Monetary Policy. Banco Central de Chile, Santiago, Chile, pp. 271–292. Levine, R., Loayza, N., Beck, T. (2000). “Financial intermediation and growth: Causality and causes”. Journal of Monetary Economics 46, 31–77. Levine, R., Renelt, D. (1992). “A sensitivity analysis of cross-country growth regressions”. American Economic Review 82, 942–963. Levine, R., Schmukler, S. (2003). “Migration, spillovers, and trade diversion: The impact of internationalization on stock market liquidity”. National Bureau of Economic Research Working Paper No. 9614. Levine, R., Schmukler, S. (2004). “Internationalization and the evolution of corporate valuations.” Mimeo. University of Minnesota (Carlson School of Management). Levine, R., Zervos, S. (1998a). “Stock markets, banks, and economic growth”. American Economic Review 88, 537–558. Levine, R., Zervos, S. (1998b). “Capital control liberalization and stock market development”. World Development 26, 1169–1184. Loayza, N., Ranciere, R. (2002). “Financial fragility, financial development, and growth”. World Bank mimeo. Love, I. (2003). “Financial development and financing constraint: International evidence from the structural investment model”. Review of Financial Studies 16, 765–791. Lucas, R.E. (1988). “On the mechanics of economic development”. Journal of Monetary Economics 22, 3–42. Luintel, K., Khan, M. (1999). “A quantitative reassessment of the finance-growth nexus: Evidence from a multivariate VAR”. Journal of Development Economics 60, 381–405. Maurer, N., Haber, S. (2004). “Related lending and economic performance: Evidence from Mexico”. Mimeo. Stanford University. McKinnon, R.I. (1973). Money and Capital in Economic Development. Brookings Institution, Washington, DC. Meier, G.M., Seers, D. (1984). Pioneers in Development. Oxford University Press, New York. Merton, R.C. (1987). “A simple model of capital market equilibrium with incomplete information”. Journal of Finance 42, 483–510. Merton, R.C. (1992). “Financial innovation and economic performance”. Journal of Applied Corporate Finance 4, 12–22. Merton, R.C. (1995). “A functional perspective of financial intermediation”. Financial Management 24, 23– 41. Merton, R.C., Bodie, Z. (1995). “A conceptual framework for analyzing the financial environment”. In: Crane, D.B., et al. (Eds.), The Global Financial System: A Functional Perspective. Harvard Business School Press, Boston, MA, pp. 3–31. Merton, R.C., Bodie, Z. (2004). “The design of financial systems: Towards a synthesis of function and structure”. National Bureau of Economic Research Working Paper Number 10620.
932
R. Levine
Miller, M.H. (1998). “Financial markets and economic growth”. Journal of Applied Corporate Finance 11, 8–14. Morales, M.F. (2003). “Financial intermediation in a model of growth through creative destruction”. Macroeconomic Dynamics 7, 363–393. Morck, R., Nakamura, M. (1999). “Banks and corporate control in Japan”. Journal of Finance 54, 319–340. Morck, R., Stangeland, D., Yeung, B. (2000). “Inherited wealth, corporate control, and economic growth: The Canadian disease”. In: Morck, R. (Ed.), Concentrated Corporate Ownership. National Bureau of Economics Research. Morck, R., Wolfenzon, D., Yeung, B. (2005). “Corporate governance, economic entrenchment and growth”. Journal of Economic Literature. In press. Morck, R., Yeung, B., Yu, W. (2000). “The information content of stock markets: Why do emerging markets have synchronous stock price movements”. Journal of Financial Economics 58, 215–260. Morgan, D. (2002). “Rating banks: Risk and uncertainty in an opaque industry”. American Economic Review 92, 874–888. Myers, S.C., Majluf, N. (1984). “Corporate financing and investment decisions when firms have information that investors do not have”. Journal of Financial Economics 13, 187–221. Neusser, K., Kugler, M. (1998). “Manufacturing growth and financial development: Evidence from OECD countries”. Review of Economics and Statistics 80, 636–646. Obstfeld, M. (1994). “Risk-taking, global diversification, and growth”. American Economic Review 84, 1310–1329. Pagano, M., Volpin, P. (2001). “The political economy of finance”. Oxford Review of Economic Policy 17, 502–519. Patrick, H. (1966). “Financial development and economic growth in underdeveloped countries”. Economic Development Cultural Change 14, 174–189. Peek, J., Rosengren, E.S. (1998). “The international transmission of financial shocks: The case of Japan”. American Economic Review 87, 495–505. Petersen, M.A., Rajan, R.G. (1997). “Trade credit: Some theories and evidence”. Review of Financial Studies 10, 661–692. Pollard, S., Ziegler, D. (1992). “Banking and industrialization: Rondo Cameron twenty years on”. In: Cassis, Y. (Ed.), Finance and Financiers in European History 1880–1960. Cambridge University Press, New York, pp. 17–38. Rajan, R.G. (1992). “Insiders and outsiders: The choice between informed and arms length debt”. Journal of Finance 47, 1367–1400. Rajan, R.G., Zingales, L. (1998). “Financial dependence and growth”. American Economic Review 88, 559– 586. Rajan, R.G., Zingales, L. (1999). “Which capitalism? Lessons from the East Asian crisis”. Journal of Applied Corporate Finance 11, 40–48. Rajan, R.G., Zingales, L. (2003). Saving Capitalism from the Capitalists. Random House, New York. Ramakrishnan, R.T.S., Thakor, A. (1984). “Information reliability and a theory of financial intermediation”. Review of Economic Studies 51, 415–432. Rioja, F., Valev, N. (2004a). “Finance and the sources of growth at various stages of economic development”. Economic Inquiry 42, 27–40. Rioja, F., Valev, N. (2004b). “Does one size fit all?: A reexamination of the finance and growth relationship”. Journal of Development Economics 74 (2), 429–447. Robinson, J. (1952). The Rate of Interest and Other Essays. MacMillan, London. Chapter “The generalization of the general theory”. Roe, M. (1994). Strong Managers, Weak Owners: The Political Roots of American Corporate Finance. Princeton University Press, Princeton, NJ. Roubini, N., Sala-i-Martin, X. (1992). “Financial repression and economic growth”. Journal of Development Economics 39, 5–30. Roubini, N., Sala-i-Martin, X. (1995). “A growth model of inflation, tax evasion, and financial repression”. Journal of Monetary Economics 35, 275–301.
Ch. 12:
Finance and Growth: Theory and Evidence
933
Rousseau, P.L. (1998). “The permanent effects of innovation on financial depth: Theory and US historical evidence from unobservable components models”. Journal of Monetary Economics 42, 387–425. Rousseau, P.L. (1999). “Finance, investment, and growth in Meiji-era Japan”. Japan and the World Economy 11, 185–198. Rousseau, P.L., Sylla, R. (1999). “Emerging financial markets and early U.S. growth”. National Bureau of Economic Research Working Paper No. 7448. Rousseau, P.L., Sylla, R. (2001). “Financial system, economic growth, and globalization”. National Bureau of Economic Research Working Paper No. 8323. Rousseau, P.L., Wachtel, P. (1998). “Financial intermediation and economic performance: Historical evidence from five industrial countries”. Journal of Money, Credit and Banking 30, 657–678. Rousseau, P.L., Wachtel, P. (2000). “Equity markets and growth: cross-country evidence on timing and outcomes, 1980–1995”. Journal of Business and Finance 24, 1933–1957. Rousseau, P.L., Wachtel, P. (2002). “Inflation thresholds and the finance-growth nexus”. Journal of International Money and Finance. Saint-Paul, G. (1992). “Technological choice, financial markets and economic development”. European Economic Review 36, 763–781. Scharfstein, D. (1988). “The disciplinary role of takeovers”. Review of Economic Studies 55, 185–199. Schumpeter, J.A. (1912). Theorie der wirtschaftlichen Entwicklung. Dunker & Humblot, Leipzig. The Theory of Economic Development translated by R. Opie. Harvard University Press, Cambridge, MA, 1934. Sirri, E.R., Tufano, P. (1995). “The economics of pooling”. In: Crane, D.B., et al. (Eds.), The Global Financial System: A Functional Approach. Harvard Business School Press, Boston, MA, pp. 81–128. Shan, J.Z., Morris, A.G., Sun, F. (2001). “Financial development and economic growth: An egg and chicken problem?”. Review of International Economics 9, 443–454. Shleifer, A., Summers, L. (1988). “Breach of trust in hostile takeovers”. In: Auerbach, A. (Ed.), Corporate Takeovers: Causes and Consequences. University of Chicago Press, Chicago, pp. 33–56. Shleifer, A., Vishny, R.W. (1996). “Large shareholders and corporate control”. Journal of Political Economy 94, 461–488. Shleifer, A., Vishny, R.W. (1997). “A survey of corporate governance”. Journal of Finance 52, 737–783. Smith, A. (1776). An Inquiry into the Nature and Causes of the Wealth of Nations. W. Stahan & T. Cadell, London. Smith, B.D. (2002). “Taking intermediation seriously”. Journal of Money, Credit, and Banking. Stein, J.C. (1988). “Takeover threats and managerial myopia”. Journal of Political Economy 96, 61–80. Stiglitz, J.E. (1985). “Credit markets and the control of capital”. Journal of Money, Credit and Banking 17, 133–152. Stiglitz, J., Weiss, A. (1983). “Incentive effects of terminations: Applications to credit and labor markets”. American Economic Review 73 (5), 912–927. Stulz, R. (1988). “Managerial control of voting rights: Financing policies and the market for corporate control”. Journal of Financial Economics 20, 25–54. Stulz, R.M. (2001). “Does financial structure matter for economic growth? A corporate finance perspective”. In: Demirgüç-Kunt, A., Levine, R. (Eds.), Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development. MIT Press, Cambridge, MA, pp. 143–188. Stulz, R.M., Williamson, R. (2003). “Culture, openness, and finance”. Journal of Financial Economics. Sussman, O. (1993). “A theory of financial development”. In: Giovannini, A. (Ed.), Finance and Development: Issues and Experience. Cambridge University Press, Cambridge, pp. 29–64. Sylla, R.E. (1998). “U.S. securities markets and the banking system, 1790–1840”. Federal Reserve Bank of St. Louis Review 80, 83–104. Tadesse, S. (2002). “Financial architecture and economic performance: International evidence”. Journal of Financial Intermediation 11, 429–454. Townsend, R.M. (1979). “Optimal contracts and competitive markets with costly state verification”. Journal of Economic Theory 21, 265–293. Weinstein, D.E., Yafeh, Y. (1998). “On the costs of a bank-centered financial system: Evidence from the changing main bank relations in Japan”. Journal of Finance 53, 635–672.
934
R. Levine
Wenger, E., Kaserer, C. (1998). “The German system of corporate governance: A model which should not be imitated”. In: Black, S.W., Moersch, M. (Eds.), Competition and Convergence in Financial Markets: The German and Anglo-American Models. North-Holland, New York, pp. 41–78. Williamson, S.D., Wright, R. (1994). “Barter and monetary exchange under private information”. American Economic Review 84, 104–123. Wright, R.E. (2002). The Wealth of Nations Rediscovered: Integration and Expansion in American Financial Markets, 1780–1850. Cambridge University Press, Cambridge, UK. Wurgler, J. (2000). “Financial markets and the allocation of capital”. Journal of Financial Economics 58, 187–214. Xu, Z. (2000). “Financial development, investment, and growth”. Economic Inquiry 38, 331–344. Zingales, L. (1994). “The value of the voting right: A study of the Milan stock exchange experience”. The Review of Financial Studies 7, 125–148.
Chapter 13
HUMAN CAPITAL AND TECHNOLOGY DIFFUSION* JESS BENHABIB AND MARK M. SPIEGEL† Economic Research, Federal Reserve Bank of San Francisco, 101 Market St., San Francisco, CA 94105, USA e-mail:
[email protected]
Contents Abstract Keywords 1. Introduction 2. Variations on the Nelson–Phelps model 3. Some microfoundations based on the diffusion model of Barro and Sala-iMartin 4. A nested specification 5. Empirical evidence 5.1. Measurement of total factor productivity 5.2. Model specification 5.3. Results 5.3.1. Base specification 5.3.2. Average human capital levels 5.3.3. Conditioning on country characteristics
6. Model prediction 6.1. Model forecasting 6.2. Negative catch-up countries
7. Conclusion References
936 936 937 940 944 946 948 948 953 954 954 956 956 959 959 960 964 965
* Very helpful comments were received from Richard Dennis, Rody Manuelli, Chris Papageorgiou, and seminar participants at LSU, USC, and the SIEPR/FRBSF Conference on Technical Change. Edmund Chiang provided excellent research assistance. The opinions in this paper are the author’s own, and do not necessarily reflect those of the Federal Reserve Bank of San Francisco, or the Board of Governors of the Federal Reserve. We thank the C. V. Starr Center at NYU for technical assistance. † Corresponding author.
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01013-0
936
J. Benhabib and M.M. Spiegel
Abstract This paper generalizes the Nelson–Phelps catch-up model of technology diffusion. We allow for the possibility that the pattern of technology diffusion can be exponential, which would predict that nations would exhibit positive catch-up with the leader nation, or logistic, in which a country with a sufficiently small capital stock may exhibit slower total factor productivity growth than the leader nation. We derive a nonlinear specification for total factor productivity growth that nests these two specifications. We estimate this specification for a cross-section of nations from 1960 through 1995. Our results support the logistic specification, and are robust to a number of sensitivity checks. Our model also appears to predict slow total factor productivity growth well. 22 of the 27 nations that we identify as lacking the critical human capital levels needed to achieve faster total factor productivity growth than the leader nation in 1960 did achieve lower growth over the next 35 years.
Keywords human capital, technology diffusion JEL classification: O4
Ch. 13:
Human Capital and Technology Diffusion
937
1. Introduction In a short paper in 1966 Nelson and Phelps offered a new hypothesis to explain economic growth. Their explanation had two distinct components. The first component postulated that while the growth of the technology frontier reflects the rate at which new discoveries are made, the growth of total factor productivity depends on the implementation of these discoveries, and varies positively with the distance between the technology frontier and the level of current productivity. Applied to the diffusion of technology between countries, with the country leading in total factor productivity representing the technology frontier, this is a formalization of the catch-up hypothesis that was originally proposed by Gerschenkron (1962). The second component of the Nelson–Phelps hypothesis suggested that the rate at which the gap between the technology frontier and the current level of productivity is closed depends on the level of human capital. This was a break with the view that human capital is an input into the production process. Nelson and Phelps make this point starkly in the concluding sentence of their paper: “Our view suggests that the usual, straightforward insertion of some index of educational attainment in the production function may constitute a gross mis-specification of the relation between education and the dynamics of production.” The catch-up or technology diffusion component of the Nelson–Phelps hypothesis raises a basic question. If a country, or a firm within an industry, has to incur costs in order to innovate, then why should it not sit back and wait for technology diffusion to flow costlessly? Modern theories of economic growth have paid a great deal of attention to the incentives for innovation and to the market structures that are necessary to sustain R&D. Inventions are typically assumed to give rise to new (often intermediate) products which generate monopoly rents over their lifetime. These rents provide the financial incentives to innovate and to cover the costs of innovation. The costs of invention typically reflect the wages or the patent incomes of researchers. The labor markets allocate workers between research and production, and in certain cases the allocation of workers across different occupations can involve decisions to acquire costly human capital. When a vintage structure is present, newer and technologically more efficient intermediate goods or production processes may coexist with older ones that remain inside the technology frontier. A critical by-product of an innovation, not captured by the monopoly rents that it generates, is the expansion of the stock of basic knowledge. This basic knowledge, freely available to all, enhances the productivity of future research, facilitates future innovations and is the source of scale effects. In the Nelson–Phelps framework, disembodied technical know-how flows from the technology leader to its followers and augments their total factor productivity. Patent protection or blueprint ownership is not explicitly postulated, and therefore an alternative mechanism must be in operation to sustain inventive activity and to prevent free-riding. A number of models have directly addressed the impact of imitation that dissipates rents on innovative activity by explicitly introducing costs of imitation. In an early investigation by Grossman and Helpman (1991, Chapter 11) [see also Helpman (1993), Segerstrom (1991)], the North, where patent protection is in effect, innovates,
938
J. Benhabib and M.M. Spiegel
and the South, where labor costs are lower, imitates at a cost. Aghion, Christopher and Vickers (1997), building on Grossman and Helpman (1991), suggest a leapfrogging model where firms can, by incurring an appropriate cost, catch-up and overtake their rivals to capture a larger share of the profits. Eaton and Kortum (1996a, 1996b) construct a model with patenting costs where patents decrease but do not eliminate the hazard of imitation. To construct an equilibrium with technology diffusion, Barro and Sala-i-Martin (1995, 1997) introduce a model where in the leading country the costs of innovation are low relative to the costs of imitation, while in the follower country the reverse is true. Basu and Weil (1998) propose a model where technological barriers to imitation in the South arise from significant differences in factor proportions between North and South, with the possible emergence of “convergence clubs”. Such differences in endowments may not provide the most “appropriate” opportunities for imitation, and fail to direct technical change towards efficient cost savings [see Acemoglu (2003)]. Technology may nevertheless flow between convergence clubs, with imitation costs rather than patent protection sustaining innovative activity within the clubs. Eeckhout and Jovanovic (2000) construct a model where imitators can implement technology only with a lag, and this implicit imitation cost means that innovators find it optimal to maintain their lead. It seems clear then that some costs of imitation and certain advantages to innovation must be present if technology diffusion is to play a role in economic growth. Therefore, underlying the Nelson–Phelps model there must be an appropriate market structure with an economic equilibrium that sustains innovative activity in the face of technology diffusion. The empirical literature on technology diffusion has been growing, despite difficulties in measurements. The survey of Griliches (1992) lends support to the view that there are significant R&D spillovers. Coe and Helpman (1995) find that R&D abroad benefits domestic productivity, possibly through the transfer of technological know-how via trade. Branstetter (2001), looking at disaggregated data, finds research spillovers across firms that are close in “technology space”. Nadiri and Kim (1996) suggest that the importance of research spillovers across countries varies with the country: domestic research seems important in explaining productivity in the US but the contribution of foreign research is more important for countries like Italy or Canada. The role of human capital in facilitating technology adoption is documented by Welch (1975), Bartel and Lichtenberg (1987) and Foster and Rosenzweig (1995). Benhabib and Spiegel (1994), using crosscountry data, investigate the Nelson–Phelps hypothesis and conclude that technology spills over from leaders to followers, and that the rate of the flow depends on levels of education. In fact a good deal of the recent empirical literature has focused on whether the level of education speeds technology diffusion and leads to growth, as suggested by Nelson and Phelps, or whether education acts as a factor of production, either directly or through facilitating technology use. [See for example, Islam (1995), Eaton and Kortum (1996a, 1996b), Temple (1998), Krueger and Lindahl (2001), Klenow and RodriguezClare (1997), Hall and Jones (1999), Bils and Klenow (2000), Duffy and Papageorgiou (2000), and Hanushek and Kimko (2000).]
Ch. 13:
Human Capital and Technology Diffusion
939
The policy implications of distinguishing between the role of education as a factor of production and a factor that facilitates technology diffusion are significant. In the former, the benefit of an increase in education is its marginal product. In the latter, because the level of education affects the growth rate of total factor productivity and technology diffusion, its benefit will be measured in terms of the sum of its impact on all output levels in the future. Following Nelson and Phelps (1966), in Benhabib and Spiegel (1994) we characterize the latter relationship through a specification that includes a term interacting the stock of human capital with backwardness, measured as a country’s distance from the technology leader. There are potentially important implications of distinguishing between different functional forms for the technology diffusion process.1 The technology diffusion process specified by Nelson and Phelps and widely used in the literature is known as the confined exponential diffusion [Banks (1994)]. An alternative diffusion process is the logistic model of technology diffusion. A priori, there appears to be no reason to favor one of these specifications over the other, and they appear to differ by very little. Nevertheless, as we demonstrate below, these specifications can have very different implications for a nation’s growth path: For the exponential diffusion process, the steady state is, for all parameterizations, a balanced growth path, with all followers growing at the pace determined by the leader nation acting as the locomotive. In contrast, the logistic model allows for a dampening of the diffusion process so that the gap between the leader and a follower can keep growing. Indeed, we demonstrate that if the human capital stock of a follower is sufficiently low, the logistic diffusion model implies divergence in total factor productivity growth rates, not catch-up. On this point, also see Howitt and Mayer-Foulkes (2002). Below we derive an empirical specification that nests these two forms of technology diffusion in a model where total factor productivity growth depends on initial backwardness relative to the stock of potential world knowledge, proxied in our model as the total factor productivity level of the leader country. We then test this specification for a cross-section of total factor productivity growth of 84 countries from 1960 through 1995. We obtain robust results supporting a positive role for human capital as an engine of innovation, as well as a facilitator of catch-up in total factor productivity. As our results favor a logistic form of technology diffusion, some countries may indeed experience divergence in total factor productivity growth. To investigate this result, we derive a point estimate for the minimum initial human capital level necessary to exhibit catch-up in total factor productivity relative to the leader nation, which is the
1 Why is technology diffusion not instantaneous? One possible explanation is that diffusion requires the transmission of information, which takes place along the lines suggested by models of epidemiology. The rate of adoption of new technologies is lowest when there are very few or very many non-users, the former because of limited contact and exposure, and the latter because the base of new users is already large. Alternatively, diffusion may be modeled as driven by a “probit” structure, where firms are heterogeneous and the timing of their adoption decisions depend on characteristics like firm size, learning costs etc. We will not address these issues here, but for a survey see Geroski (1999).
940
J. Benhabib and M.M. Spiegel
United States in our sample. The point estimate in our favored specification indicates that an average of 1.78 years of schooling was required in 1960 to allow convergence in total factor productivity growth with the United States. Under this criterion, we identify 27 countries in our sample that our point estimates predict will exhibit slower total factor productivity growth than the United States. Our data shows that over the next 35 years, 22 of these 27 countries did indeed fall farther behind the United States in total factor productivity, while the remaining bulk of the nations in our sample exhibited positive catch-up in total factor productivity. While this result is not a formal test of our model, its ability to correctly identify countries that would subsequently exhibit slower total factor productivity growth than the United States is reassuring. We then repeat our exercise using 1995 figures to identify the set of nations that are still falling behind in total factor productivity growth. Because the United States had higher education levels in 1995, we estimate a higher threshold level for total factor productivity growth convergence with the United States. Our estimate is that 1.95 average years of schooling in the population over the age of 25 is necessary for faster total factor productivity growth than the leader nation. Fortunately, the higher overall education levels achieved by most countries over the past 35 years left few countries falling the threshold levels in education to achieve catch-up in growth rates. We identified only four countries that were still below the threshold in 1995: Mali, Mozambique, Nepal, and Niger. With the exception of these four nations, our results indicate that most of the world is not in a permanent development trap, at least in terms of total factor productivity growth. Nevertheless, it should be pointed out that catch-up in total factor productivity and in growth rates is not a guarantee of convergence in per capita income, as nations must also be successful in attracting physical capital to achieve the latter goal. The remainder of the paper is divided into five sections. Section 2 introduces the exponential and logistic specifications of the Nelson–Phelps model and examines their steady-state implications. Section 3 compares the diffusion models with that of Barro and Sala-i-Martin (1997). Section 4 derives a non-linear growth specification that nests the exponential and logistic technology diffusion functional forms. Section 5 estimates this model using maximum likelihood for a cross-section of countries. Section 6 uses the point estimates from our estimation to identify nations that are predicted to fail to exhibit divergence in total factor productivity growth in 1960 and 1995. Lastly, Section 7 concludes.
2. Variations on the Nelson–Phelps model We will examine the implications of two types processes often studied in the context of disaggregated models of technology diffusion [Banks (1994)]. We can express the original Nelson–Phelps model of technology diffusion as follows: Am (t) A˙ i (t) (2.1) = g Hi (t) + c Hi (t) −1 Ai (t) Ai (t)
Ch. 13:
Human Capital and Technology Diffusion
941
where Ai (t) is the TFP, gi (Hi (t)) is the component of TFP growth that depends on (t) the level of education Hi (t) in country i and c(Hi (t))( AAmi (t) − 1) represents the rate of technology diffusion from the leader country m to country i. We assume that ci (·) and gi (·) are increasing functions. The level of education Hi (t) affects the rate at which (t) the technology gap ( AAmi (t) − 1) is closed. If the ranking of gi (Hi (t)) across countries do not change, or if Hi ’s are constant, a technology leader will emerge in finite time with gm = g(Hm (t)) > g(Hi (t)) = gi . After that the leader will grow at rate gm and the followers will fall behind in levels of TFP until the point at which their growth rate will match the leader’s growth rate gm . This can be seen from the solution of the above equation when Hi ’s are constant:2 Ai (t) = Ai (0) − ΩAm (0) e(gi −ci )t + ΩAm (0)egm t (2.2) where ci = c(Hi ), gi = g(Hi ) and Ω=
ci > 0. c i − gi + gm
It is clear, since gm > gi , that lim
t→∞
Ai (t) = Ω. Am (t)
This is, for all parameterizations, a world balanced growth path with the leader acting as the “locomotive”. Technology diffusion and “catch-up” assures that despite scale effects and educational differences, all countries eventually grow at the same rate.3 The technology diffusion and catch-up processes outlined above are also known as the confined exponential diffusion process [see Banks (1994)]. An alternative formulation that is similar in spirit is the logistic model of technology diffusion [see Sharif (1981)]. It is given by A˙ i (t) Ai (t) = g Hi (t) + c Hi (t) 1 − Ai (t) Am (t) Ai (t) Am (t) = g Hi (t) + c Hi (t) (2.3) −1 . Am (t) Ai (t)
2 The general solution when H ’s are not constant is given by: i t
Ai (t) = Ai (0)e− 0 (g(Hi (s))−c(Hi (s))) ds t τ τ 1 · 1+ c Hi (τ ) Am (0)e 0 g(Hm (ζ )) dζ e 0 (g(Hi (ξ ))−c(Hi (ξ ))) dξ dτ . Ai (0) 0 3 Note however that in transition, the higher is initial A (0), the smaller is the technology gap to the leader i and therefore the slower is the growth. This negative dependence on initial conditions is similar to standard convergence results in the neoclassical growth model, but the logic of catch-up is different.
942
J. Benhabib and M.M. Spiegel
The difference of the dynamics under the logistic model of technology diffusion and the confined exponential one is due to the presence of the extra term ( AAmi (t) (t) ). This term acts to dampen the rate of diffusion as the distance to the leader increases, reflecting perhaps the difficulty of adopting distant technologies. As shown by Basu and Weil (1998), the frontier technology may not be immediately “appropriate” for the follower if differences in factor proportions between leader and follower are large. We may observe convergence clubs, as documented by Durlauf and Johnson (1995), from which follower countries can break out only by investing in physical and human capital. Catch-up therefore may be slower when the leader is either too distant or too close, and is fastest at intermediate distances.4 If we assume, as before, that Hi ’s (and therefore, ci ’s and gi ’s) are constant such that Hm > Hi , and therefore that c(Hm ) > c(Hi ), then the solution to the logistic technology diffusion equation is given by5,6,7 Ai (t) =
Ai (0)e(gi +ci )t 1+
ci Ai (0) (ci +gi −gm )t Am (0) (ci +gi −gm ) (e
− 1)
> 0.
(2.4)
This equation can be written as Ai (t) =
Am (0)egm t Am (0) e−(ci +gi −gm )t Ai (0) − (ci +gcii−gm ) +
ci (ci +gi −gm )
(2.5)
so that in the limit,
c i + gi − gm ci Ai (t) Ai (0) lim = t→∞ Am (t) Am (0) 0
if ci + gi − gm > 0, if ci + gi − gm = 0,
(2.6)
if ci + gi − gm < 0.
4 An alternative view of technology adoption through diffusion that follows a logistic pattern borrows from epidemiology. The rate of adoption in a fixed population may depend on the rate of contact between adopters and hold-outs (those that are infected and those that are healthy). The adoption rate is highest when there are an equal number of both types, and lower when there is either a small or a large proportion of adopters. Also observing the successes and implementation errors of the first adopters, together with the competitive pressures that first adopters create, may result in a speeding up of adoption rates. See Mansfield (1968). 5 Provided that (c + g − g ) = 0. If (c + g − g ) = 0, then the equation reduces to exponential form m m i i i i Ai (t) = Ai (0)e(gi +ci )t . 6 The general solution where H ’s are functions of time can be computed by defining B = (A )−1 and i i i transforming the logistic form into the confined exponential. After some computations, the general form can be obtained as
Ai (t) =
t Ai (0)e 0 (g(Hi (s))+c(Hi (s))) ds
τ τ . 1 + Ai (0)( 0t c(Hi (τ ))((Am (0)−1 )e− 0 g(Hm (ζ )) dζ )e 0 (g(Hi (ξ ))+c(Hi (ξ ))) dξ dτ )
ci 7 A (t) > 0 because when c + g − g = 0, (ci +gi −gm )t − 1) > 0. m i i i ci +gi −gm (e
Ch. 13:
Human Capital and Technology Diffusion
943
Equation (2.6) implies that in the case of the logistic diffusion model, the steady state growth relationship will depend on the relative magnitude of the catch-up rate and the difference in the growth rate due to innovation, gm − gi . If the catch-up rate exceeds the differential growth rate solely due to educational differences between the leader and follower, that is if c(Hi ) + g(Hi ) − g(Hm ) > 0, then the leader will have a locomotive effect and pull the followers along. In such a case growth rates will converge. However, if the education level of a follower is so low that c(Hi ) + g(Hi ) − g(Hm ) < 0, then the follower will not be able to keep up, growth rates will diverge, and the income ratio of the follower to the leader will go to zero. This highlights the critical role of the type of technology diffusion process and its interaction with education in fostering economic growth: a country with a low level of education may still keep within the gravitational pull of the technology leader, provided that the level of education is high enough to permit sufficient diffusion. If technology diffusion is of the logistic type, countries with educational levels that are too low will get left behind and we may observe the phenomenon of “convergence clubs”. Escaping from the lower “club” is nevertheless possible through investments in human capital, as discussed by Basu and Weil (1998).8 The implications of logistic versus exponential technology diffusion for economic growth can therefore be quite divergent. Note that we can append the Nelson–Phelps framework, either in the logistic or the confined exponential form, to the Romer (1990) model by adding the catch-up term the research sector producing the blueprints A. The marginal product of H in the research sector will now reflect an effect from the catch-up term, and increase the allocation of H towards the research sector away from production or leisure. If, as in Romer, we assume that H is constant while knowledge, A, is accumulated, and also assume that goods use labor but not H , we may focus on the allocation of H to imitation through catch-up or to innovation. Adopting a linear specification with g(Hi ) = gHi , c(Hi ) = cHi , the marginal product of H in innovation is given by gAi (t) while in imitation, for the con(t) − 1). These marginal products are independent fined exponential case, it is cAi (t)( AAmi (t) of Hi so we may have a bang–bang solution, with all of Hi allocated towards catch-up and imitation up to a threshold, and to innovation otherwise.9 In what follows we will, for the time being, abstract from issues regarding the allocation of Hi , and assume that all of it enters both imitation and catch-up as a non-excludable public good.
8 We should note that c(H ) may also depend on barriers to innovation as in Parente and Prescott (1994), so i that in fact we have c(Hi , X), where X represents the level of barriers. 9 A further consideration is the allocation of resources between imitative and innovative uses, where the efficient allocation changes as the distance to the technology frontier narrows. The market allocation may differ from the efficient allocation due to a variety of factors, and policy interventions may improve welfare. See Acemoglu, Aghion and Zilibotti (2002).
944
J. Benhabib and M.M. Spiegel
3. Some microfoundations based on the diffusion model of Barro and Sala-i-Martin To set the stage first we express the confined exponential and logistic growth equations discussed above in stationary variables by defining B(t) =
Ai (t) −gm t e A∗ (0)
(3.1)
for all i. Then, for the logistic case, we have B˙ = c(Hi )(1 − B) + g(Hi ) − g(Hm ), B B˙ = c(Hi ) + g(Hi ) − g(Hm ) B − c(Hi )B 2 . If Hi ’s are fixed the solution is c +g −gm −1 c i + gi − gm ci + gi − gm A∗ (0) − i ci t i B(t) = 1+ −1 e . ci ci A(0)
(3.2)
(3.3)
So if ci + gi − gm > 0, c i + gi − gm , ci
lim B(t) =
t→∞
while if ci + gi − gm < 0, limt→∞ B(t) = 0.10 Note from Equation (3.2) that in the latter case where ci + gi − gm < 0, there is no steady state with B > 0. In the confined exponential case B˙ = c(Hi ) B −1 − 1 + g(Hi ) − g(Hm ), B B˙ = c(Hi ) − c(Hi ) + g(Hm ) − g(Hi ) B.
(3.4)
Since c(Hi ) + g(Hm ) − g(Hi ) > 0, it is clear from (3.4) that there exists a stable steady c(Hi ) . state at B = c(Hi )+g(H m )−g(Hi ) In the Barro and Sala-i-Martin (1997) model, the North, where innovation is cheap, is the leader. It innovates by introducing new intermediate goods, and receives no diffusion through imitation from the South. As in a typical growth model of the Romer type, it grows at a constant rate γ . The South introduces new intermediate goods through imitation. In both countries the production of final goods is given by: Yi = Ai (Li )
1−α
Ni
(Xij )α ,
i = 1, 2
j =1
10 In the case c + g − g > 0, B(t) should (if the assumption that H ’s are constant holds) exhibit the m i i i
S-shaped logistic diffusion.
Ch. 13:
Human Capital and Technology Diffusion
945
where the North is country 1 and the South is country 2, so that N1 > N2 . The profits of the j th intermediate goods producer is given by π2j = (P2j − 1)X2j where P2j is the price of the intermediate good in terms of the final good in the South. The cost of imi2 tation in the South is v2 N N1 . In a symmetric equilibrium investment in R&D is given by v2 N˙ 2 = Y2 − C2 − N2 X2
(3.5)
where the left-hand side is the cost of introducing a new intermediate good through imitation, and the right-hand side is income minus consumption minus the cost of operating the existing intermediate goods (since Xi2 = X2 for all i). Barro and Sala-i-Martin show that in equilibrium X2 and Y2 /N2 are constants.11,12 For simplicity of exposition we will also assume a constant consumption propensity out of income, so that C2 = µ(Y2 − N X2 ), so that 1 1 Y2 N˙ 2 = − X2 (1 − µ) ≡ P N2 v 2 N2 v2 and 1 B˙ = P −γ B v2
(3.6)
2 13 where B = N N1 . Barro and Sala-i-Martin assume that σ N2 v2 = η ≡ ηB σ , σ > 0. N1
Imitation costs are higher, the closer the follower is to the leader. We can now assume that η depends negatively on human capital, so that the cost of imitation declines with H . Introducing this specification into (3.6) we get B˙ = η−1 B 1−σ P − γ B = B η−1 P B −σ − γ −1/σ . Therefore this specification of imitawhich has a stable steady state at B = ( ηγ P ) tion costs yields the same qualitative conclusions as the confined exponential diffusion
11 In particular, X = L (A )1/(1−α) (α)2/(1−α) and Y /N = (A )1/(1−α) α (2α)/(1−α) L where L is, 2 2 2 2 2 2 2 2
for simplicity, the constant the labor supply in the South. 12 In Barro and Sala-i-Martin (1997), consumption growth depends on the interest rate, which reflects the
value of the stream of profits divided by the cost of imitation. Since the cost of imitation depends on N2 /N1 , the dynamic system is two-dimensional in N2 /N1 and C2 /N1 . For details, see Barro and Sala-i-Martin (1997). 13 To see that Y − N X corresponds to income note that in a symmetric equilibrium (Y − 2 2 2 2 N2 N2 0 P2j X2j dj ) + 0 π2j dj = Y2 − N2 (P2 X2 − π2 ). Thus we must show that P2 X2 − π2 = X2 where 1/(1−α) , X = L (A )1/(1−α) (α)2/(1−α) π2 is profits. Since in equilibrium π2 = (α −1 − 1)(α 2 AL1−α 2 2 2 2 )
1/(1−α) − (α −1 − 1)(α 2 AL1−α )1/(1−α) = and P2 = α −1 , we have P2 X2 − π2 = α −1 (α 2 AL1−α 2 ) 2 1/(1−α) = X . (α 2 AL1−α 2 2 )
946
J. Benhabib and M.M. Spiegel
used by Nelson and Phelps: the leader acts as the engine of growth pulling the followers along. We now modify the imitation costs to correspond to the case of logistic technology diffusion. Let v2 = η(1 − B)−1 where again v2 is increasing in B. Now the diffusion equation becomes B˙ = η−1 P (1 − B)B − γ B = η−1 P − γ B − η−1 P B 2
(3.7)
which has the same logistic structure as (3.2). In particular, there is a positive steady −1 state B = η η−1P P−γ only if η−1 P > γ . Otherwise B converges to 0, and there is no catch-up. More generally since the non-zero steady state is given by v2 (B ∗ ) = B∗
P γ
, if
v2 (0) > there will not be a positive steady state > 0, but the steady state B = 0 will be stable.14,15 If η is decreasing in H so that imitation costs decline with human capital, for sufficiently low levels of H we may have η−1 P < γ , and the South will never catch up in growth rates. In such circumstances there may be incentives to accumulate human capital. If however there are market imperfections in the accumulation of capital, or if H mostly provides external effects, there may not exist sufficient market incentives for the accumulation of H , so that subsidies to education may be necessary to improve growth and welfare. P γ
4. A nested specification We can also, for purposes of estimation, specify a diffusion process that nests the logistic and confined exponential diffusion processes. Using the definition of B given in (3.1), we can modify (3.2) as c(Hi ) B˙ = 1 − B s + g(Hi ) − g(Hm ), B s
(4.1)
14 If on the other hand, we adopt the confined exponential (ν (B) = η(B −1 − 1)) or the Barro and Sala-i2 Martin specification (ν2 (B) = ηB σ ), then ν2 (0) = 0, so that the diffusion rate approaches infinity and the imitation costs go to zero when N2 and B tend to zero, a strong and unlikely assumption. 15 In Barro and Sala-i Martin’s more general model consumption growth, given by the Euler equation, de-
pends on the rate of return on intermediate goods, which varies through time with the distance to the frontier. C At a steady state with B > 0, we obtain N2 = 1+α α π2 − γ v2 (B) where π2 is the profit rate. Plugging this 2
2 where ρ is the discount rate and θ −1 is the intertemporal into (3.5) and simplifying, we get v2 (B) = θ γ +ρ consumption elasticity. For the Barro and Sala-i Martin specification, v2 (B) = ηB σ , v2 (0) = 0. If however π2 , no positive steady state B exists. This is likely if v2 also depends (inversely) on human capital v2 (0) > θ γ +ρ and if human capital levels are low.
π
Ch. 13:
Human Capital and Technology Diffusion
947
c(Hi ) + sg(Hi ) − sg(Hm ) c(Hi ) s+1 B˙ = B− B , s s Bs c(Hi ) + sg(Hi ) − sg(Hm ) ˙ B 1− B= s 1 + s(gi −gm )
(4.2) (4.3)
ci
with s ∈ [−1, 1]. Note that if s = 1, this specification collapses to the logistic, and if s = −1, it collapses to the confined exponential.16 In its general form this is a Bernoulli equation, whose solution, when Hi and Hm are constants so that ci = c(Hi ), gm = g(Hm ), gi = g(Hi ), is given by: B(t) =
1+ 1 + ((1 +
s(gi −gm ) ci
s(gi −gm ) )B(0)−s ci
1/s
− 1)e−(ci +s(gi −gm ))t
.
(4.4)
Since the leader has more human capital, Hm > Hi , we have gm > gi . It follows that if either ci + s(gi − gm ) > 0, or if s < 0, s(gi − gm ) 1/s lim B(t) = 1 + , t→∞ ci 17 m) while if (1 + s(gi c−g ) < 0, and s > 0, limt→∞ B(t) = limt→∞ AAmi (t) (t) = 0. In the i latter case, as noted in the previous section, the South never catches up and growth rates diverge.18
16 See Richards (1959). 17 If c and g vary with time because H changes with time, (4.1) is the classic Bernoulli equation which we i i i
can write as: B˙ = f (t)B + g(t)B s+1 where f (t) =
c(Hi (t)) c(H (t)) + g(Hi (t)) − g(Hm (t)) and g(t) = − si as in Equation (4.2). The solution is: s
−1/s B(t) = Ceφ(t) + seφ(t) eφ(τ ) g(τ ) dτ where φ(t) = s f (τ ) dτ and C is an integration constant such that C −1/s = B(0). 18 When s → 0, the diffusion process converges to the Gompertz growth model: g − gm s(gi − gm ) 1/s B = lim 1 + exp −ek−ci t = exp i exp −ek−ci t , ci ci s→0 exp( gi −gm ) ci B˙ = ci B ln B
(4.5)
(4.6)
gi −gm g −g − ln(B(0)). So limt→∞ B = exp( i c m ) > 0. To see this note that, using L’Hopital’s ci i g −g exp( i c m ) i ). rule, the right side of Equation (4.3) collapses to ci B ln( B
where ek =
948
J. Benhabib and M.M. Spiegel
To test this nested specification empirically we can specify it as: Ait s c c ait = g + hit − hit , s s Amt where ait is the growth of TFP for country i, hit is its initial or average human capital and Ait /Amt is the ratio of the country’s TFP to that of the leader. Note again that this specification nests the logistic (s = 1) and exponential (s = −1) models. As discussed above, the values of c, g and s will determine whether a country will converge to the growth rate of the leader or whether the growth rates will diverge. In particular, for our linear specification c(hit ) = ci = chit , g(hit ) = gi = ghit and g(hmt ) = gm = ghmt , “the catch-up condition” for the growth rate of a country to converge to the growth rate of the leader becomes (for s ∈ (0, 1]): c∗ = 1 +
hmt c > . sg hit
(4.7)
Countries for which hmt /hit > c∗ will not converge to the leader’s growth rate unless they invest in their human capital to reverse this inequality.19
5. Empirical evidence 5.1. Measurement of total factor productivity Data for real income and population growth were obtained from the Penn World Tables, version 6.1. Data for human capital, which is proxied by average years of schooling in the population above 25 years of age, was obtained from the updated version of the Barro and Lee (1993) data set. Our sample consists of 85 countries with data for the period 1960–1995. We estimate this sample both as a cross-section of 35 years of growth and as a panel of five-year growth rates. Physical capital stocks were calculated according to the method used in Klenow and Rodriguez-Clare (1997). Initial capital stocks are calculated according to the following formula I /Y K = (5.1) Y γ +δ+n 1960
where I /Y is the average share of physical investment in output from 1960 through 2000, γ represents the average rate of growth of output per capita over that period, n represents the average rate of population growth over that period, and δ represents the rate of depreciation, which is set equal to 0.03. Given initial capital stock estimates, the 19 As noted earlier however the catch-up coefficient c(h ) may depend on other institutional factors in addiit
tion to human capital, like barriers to innovation as in Parente and Prescott (1994). In such a case we may want to modify the catch-up coefficient to ci = αi chit where αi reflects country specific barriers to innovation.
Ch. 13:
Human Capital and Technology Diffusion
949
capital stock of country i in period t satisfies Kit =
t
(1 − δ)t−j Iij + (1 − δ)t K1960 .
(5.2)
j =0
Total factor productivity growth was estimated from a constant returns to scale Cobb– Douglas production function with the capital share set at 1/3 and the labor share set at 2/3.20 For country i in period t 1 2 ait = yit − kit − lit (5.3) 3 3 where ait represents the log of total factor productivity, yit represents the log of real output, kit represents the log of the physical capital stock, and lit represents the log of the population. Total factor productivity estimates for 1960 and 1995, as well as estimates of average annual growth in total factor productivity over the period are shown in Table 1. The results seem pretty intuitive, as the Asian Tiger countries, including Taiwan, Singapore, Korea, Hong Kong and Thailand, lie notably at or near the top in terms of total factor productivity growth, while the five countries exhibiting the lowest growth in total factor productivity are Mozambique, Niger, Central African Republic, Nicaragua, and Zambia. All of these countries experienced negative total factor productivity growth over the sample period, as did Mali, Senegal, Venezuela, Togo and Cameroon. Out of this group of ten negative total factor productivity growth countries, only Venezuela’s appearance is surprising, and that can probably be attributed to its buildup of physical capital for oil production. In the case of the five highest total factor productivity growth countries, our results would no doubt differ slightly if our sample included the Asia crisis of 1997. Nevertheless, the set of countries exhibiting high total factor productivity growth seems intuitive as well. A simple scatter plot of initial human capital levels and subsequent total factor productivity growth over the estimation period is shown in Figure 1. The raw correlation between these two variables is clearly positive, suggesting that nations with larger initial human capital stocks tend to exhibit higher total factor productivity growth holding all else constant. There are a number of interesting outliers. The Asian tiger nations are noteworthy as nations that exhibited fast total factor productivity growth and began the estimation period with relatively stocks of initial human capital.21 On the other hand, 20 Gollin (2002) estimates that the share of labor lies between 0.65 and 0.80 for a cross-section of world
economies. Keller (1998) estimated TFP with both the factor shares used above and the capital and labor shares set equal to one-half and obtained similar ordinal rankings of total factor productivity levels across countries. 21 It is unfortunate that our sample ends in 1995, because the Asian “tiger” nations suffered large declines in the 1997 crisis. However, we confirmed that total factor productivity growth of these nations was still exceptionally high for the Asian tiger nations for which longer 39 year data from 1960 to 1999 was available. This included all of the tigers except Singapore.
950
J. Benhabib and M.M. Spiegel Table 1 Total factor productivity estimates (1960–1995)
Country
Mozambique Niger Central African Rep. Nicaragua Zambia Mali Senegal Venezuela Togo Cameroon Tanzania Bolivia Honduras El Salvador Guyana Peru Argentina Uganda South Africa Jamaica Philippines Costa Rica Bangladesh Jordan New Zealand Uruguay Nepal Malawi Algeria Ghana Guatemala Switzerland Kenya Mexico Papua New Guinea Iran Lesotho Trinidad and Tobago Fiji Ecuador Sweden Dominican Rep. United Kingdom Canada
log TFP1960
log TFP1995
Average annual log growth of TFP (1960–1995)
0.5010 0.2045 0.4180 0.4487 −0.2912 −0.1092 0.3209 1.0141 −0.1249 0.3181 −1.0572 0.3817 0.1513 0.7495 0.0168 0.4039 0.9538 0.0519 0.8463 0.2297 0.2176 0.6131 −0.0997 0.4289 1.1840 0.8978 −0.3250 −0.7672 0.3615 −0.2121 0.5197 1.2467 −0.2842 0.6282 0.3175 0.3787 −0.4715 0.8535 0.3940 0.1191 1.0855 0.2220 1.1090 1.1711
−0.0353 −0.2983 −0.0791 0.0546 −0.5857 −0.2677 0.1634 0.9306 −0.1917 0.2649 −1.0181 0.4642 0.2597 0.8820 0.1989 0.6054 1.1675 0.2721 1.0689 0.4554 0.4506 0.8480 0.1442 0.6773 1.4505 1.1733 −0.0416 −0.4742 0.6622 0.0893 0.8215 1.5526 0.0390 0.9549 0.6532 0.7390 −0.1054 1.2695 0.8118 0.5526 1.5350 0.6859 1.5778 1.6541
−0.0153 −0.0144 −0.0142 −0.0113 −0.0084 −0.0045 −0.0045 −0.0024 −0.0019 −0.0015 0.0011 0.0024 0.0031 0.0038 0.0052 0.0058 0.0061 0.0063 0.0064 0.0064 0.0067 0.0067 0.0070 0.0071 0.0076 0.0079 0.0081 0.0084 0.0086 0.0086 0.0086 0.0087 0.0092 0.0093 0.0096 0.0103 0.0105 0.0119 0.0119 0.0124 0.0128 0.0133 0.0134 0.0138
Ch. 13:
Human Capital and Technology Diffusion
951
Table 1 (Continued) Country
Australia Denmark Paraguay Turkey Colombia Netherlands Zimbabwe United States Sri Lanka Finland Iceland Chile India Panama France Ireland Belgium Syria Brazil Greece Austria Norway Italy Israel Pakistan Spain Mauritius Portugal Indonesia Barbados Malaysia Romania Japan Botswana Cyprus Thailand Hong Kong Rep. of Korea Singapore Rep. of China, Taiwan
log TFP1960
log TFP1995
Average annual log growth of TFP (1960–1995)
1.1472 1.1227 0.4728 0.4371 0.4648 1.0327 −0.2344 1.3257 0.0648 0.8676 0.9602 0.6381 −0.2360 0.2486 0.9176 0.8202 0.9147 0.1391 0.2618 0.5097 0.8583 0.8808 0.8291 0.7494 −0.4390 0.6153 0.6394 0.4739 −0.1621 0.5475 0.2549 −0.3987 0.5632 −0.1326 0.3582 −0.3058 0.4578 −0.0429 0.1202 0.1046
1.6339 1.6215 0.9894 0.9546 0.9855 1.5617 0.2948 1.8626 0.6074 1.4237 1.5301 1.2141 0.3458 0.8324 1.5088 1.6031 1.5555 0.7957 0.9204 1.1877 1.5445 1.5879 1.5379 1.4757 0.3175 1.4203 1.4829 1.3254 0.7056 1.4540 1.1852 0.5327 1.5851 0.9935 1.5217 0.9102 1.8604 1.3646 1.6285 1.6140
0.0139 0.0143 0.0148 0.0148 0.0149 0.0151 0.0151 0.0153 0.0155 0.0159 0.0163 0.0165 0.0166 0.0167 0.0169 0.0182 0.0183 0.0188 0.0188 0.0194 0.0196 0.0202 0.0202 0.0163 0.0216 0.0230 0.0241 0.0243 0.0248 0.0259 0.0266 0.0266 0.0292 0.0322 0.0332 0.0347 0.0401 0.0402 0.0431 0.0431
952
J. Benhabib and M.M. Spiegel
Figure 1. TFP growth vs initial human capital.
there are a number of countries that exhibited total factor productivity declines that began the period with exceptionally low levels of initial human capital, including Mali, Niger, Togo, Mozambique, and the Central African Republic.
Ch. 13:
Human Capital and Technology Diffusion
953
5.2. Model specification As discussed above, the following non-linear cross-sectional specification nests the exponential and logistic functional forms of technology diffusion Ai s c c hi − hi ai = b + g + (5.4) + εi s s Am where ai represents the average annual growth rate in TFP of country i, hi represents the log of country i’s stock of human capital, Ai represents the level of country i’s stock of TFP, Am represents the level of TFP of the leader nation, and εi is an i.i.d. disturbance term. The coefficients to be estimated represent b, g + cs , − cs , and s respectively. We are agnostic as to whether it is appropriate to include the constant term b. This term could be interpreted as exogenous technological progress that is independent of human capital and technology diffusion. It is difficult to envision any type of technological progress that would be common across our sample and completely independent of the levels of national human capital. In the case where “accidental technological progress” truly does take place, it is far more likely that it would appear in our error term as it would be confined to specific nations within our sample. Nevertheless, we report our estimation results both without and with the constant terms included as a measure of their robustness. Our model nests two alternative hypotheses. First, we have our Nelson–Phelps type model of technology diffusion, dependent on human capital and technological backwardness, that is of the confined exponential type. As we noted above, this model would correspond to the above specification with s equal to −1. Second, we have our logistic specification for the technology diffusion process, which would correspond to s being equal to 1. We therefore estimate the above nested model to let the data determine the appropriate value of s. Because our model is non-linear, we cannot use the differenced panel estimators for cross-country growth regressions that have become popular in the literature [e.g. Caselli, Esquivel and Lefort (1996), Easterly, Loayza and Montiel (1997), and Benhabib and Spiegel (2000)]. Instead, we estimate the nested specification above in a cross-sectional sample of long-term growth using maximum likelihood. In order to minimize problems with endogeneity, we use initial values for human capital stocks and initial total factor productivity. As we are comparing these initial values to the nations’ subsequent growth experiences over the next 35 years, endogeneity issues are unlikely to be a problem. We also conduct a number of robustness checks. First, there is a concern about the quality of initial human capital values as a proxy of the human capital stock available over the estimation period. Recall that our specification implies that human capital is a measure of a nation’s capacity to conduct innovation activity (accounted by the first term in the specification), and technology adoption from abroad (captured by the second term in the specification). However, many of the nations in our sample exhibited dramatic growth in their human capital stocks over this period, as measured by average years of schooling. A number of nations, including Nepal, Togo, Iran, Ghana, Syria,
954
J. Benhabib and M.M. Spiegel
and the Central African Republic, actually had more than a five-fold increase in their average years of schooling in the population over the age of 25. This implies that the initial stocks of human capital in 1960 may poorly represent the stocks of human capital available to a nation later on in the sample period. We therefore also report results using average human capital levels over the estimation period.22 However, this measure is likely to suffer more from endogeneity issues than initial human capital levels, as a nation’s financial ability to increase the average human capital levels of its citizens is likely to be increasing in its rate of output and total factor productivity growth. Fortunately, as we demonstrate below, our results are fairly robust to either measure of the stock of human capital. Second, since we are estimating a cross-section, we are unable to condition on country-specific fixed effects. In response, we further examine the robustness of our results to the introduction of a number of conditioning variables. Using data obtained from Sachs and Warner (1997), we introduce a number of geo-political characteristics, including a Sub-Saharan Africa dummy, a dummy for countries that are not landlocked, a dummy for tropical countries, a dummy for initial life expectancy, a dummy for ethnolinguistic fractionalization, and a dummy for openness over the estimation period. 5.3. Results 5.3.1. Base specification Our results with initial stocks of human capital are shown in Table 2. Our base specification is reported in Model 1. It can be seen that the coefficient on human capital, which represents (g + c/s) in the specification above, enters significantly with a positive coefficient in log levels at a 5 percent confidence level, consistent with the notion of human capital as a facilitator of own innovation predicted by the theory. The next term represents the coefficient on the catch-up term, −(c/s) in the above specification. This term enters as predicted with a negative and statistically significant sign at a 5 percent confidence level. Finally, our point estimate of s is equal to 2.304. This number is not significantly different from 1, but is significantly greater than 0 at a ten percent confidence level. These results therefore favor the logistic specification, suggesting that there is some initial human capital level below which a country would fall farther and farther behind the leader national in total factor productivity over time. We investigate this possibility in more detail below. One disappointing result in our base specification is that our point estimate for human capital lies below that of the catch-up term in absolute value. This implies that our point estimate for g is negative, which is implausible. However, this point estimate is insignificantly different from 0 and does include positive values for any standard
22 Average human capital levels are calculated as the simple averages of beginning (1960) and ending (1995)
human capital levels.
Ch. 13:
Human Capital and Technology Diffusion
955
Table 2 Regression results: log H1960 Model 1
Model 2
Model 3
Model 4
C
0.0083∗∗ (0.0016)
–
0.0085∗∗ (0.0016)
–
ln(H1960 )
0.0080∗∗ (0.0019)
0.0116∗∗ (0.0016)
0.0100∗∗ (0.0023)
0.0134∗∗ (0.0025)
−0.0086∗∗ (0.0032)
−0.0085∗∗ (0.0039)
−0.0089∗∗ (0.0036)
−0.0072∗∗ (0.0025)
2.304∗ (1.405)
3.164∗ (1.892)
1
1
84
84
84
84
log likelihood
264.5
252.4
263.9
263.9
Wald P -value
0.00
0.00
0.00
0.00
TFP s ln(H1960 ) · TFPmi
s
Number of observations
Note: Estimation by maximum likelihood with standard errors is presented in parentheses. ∗∗ Statistical significance at the 5% confidence level. ∗ Statistical significance at the 10% confidence level.
confidence level. Nevertheless, the negative point estimate does become a problem for our data exploration. In particular, using the negative point estimate for g precludes the existence of a positive critical human capital stock below which catch up in total factor productivity cannot occur. As discussed above, the problem with the specification of Model 1 is that our theory does not call for a constant term independent of human capital to account for total factor productivity growth. Consequently, Model 2 repeats our base specification with the constant term excluded. It can be seen that our qualitative results are robust to the exclusion of a constant term. Human capital in log levels again enters significantly with a positive coefficient at a 5 percent confidence level, while the catch-up term is again significantly negative at a 5 percent confidence level, as predicted by the theory. Our point estimate of s is a little higher, at 3.164, but as before we cannot reject the hypothesis that s is equal to 1 at standard confidence levels, although we again reject the hypothesis that s is less than or equal to 0 at a 10 percent confidence level. Moreover, it can be seen that our point estimate for g is positive with this specification, allowing us to calculate a critical human capital stock below which catch-up in growth rates will not occur. Models 3 and 4 repeat our estimation with and without a constant term, with s constrained to equal 1. This results in a linear specification and provides a robustness check
956
J. Benhabib and M.M. Spiegel
of the coefficients obtained in our non-linear specification. It can be seen that our point and standard error estimates are very close to those obtained with s unconstrained. Both with and without a constant term, human capital enters significantly with a positive coefficient in log levels at a 5 percent confidence level. Moreover, the catch-up term coefficient is again negative and significant at a 5 percent confidence level, as predicted. These results suggest that our findings are not dependent on the non-linear estimation of s to obtain coefficient estimates consistent with the notion of human capital playing a positive role in facilitating both innovation and catch-up. 5.3.2. Average human capital levels Our first set of robustness checks repeats our estimation using average levels of human capital over the estimation period rather than initial human capital values.23 As discussed above, we do this to address the concern that some nations’ stocks of human capital changed dramatically over the estimation period, and therefore that initial human capital values may be relatively noisy indicators of the average levels of human capital over the estimation period that determined their TFP growth. The results incorporating this change are shown in Table 3. It can be seen that our qualitative results are fairly robust. Average human capital levels enter positively and significantly, as predicted, at a 5 percent confidence level, as do the coefficient estimates for the catch-up term. The magnitudes of these coefficients are similar to those obtained with initial human capital stocks, but they are both somewhat larger in absolute value. This increase is interesting because average measured human capital levels are larger than initial human capital levels, as all nations experienced some increase in average years of schooling over the estimation period. Our estimates of s in Models 1 and 2 are very close to 1, which would again favor our logistic specification, but the large standard errors associated with our estimates of s leave them insignificantly different from 0 at standard confidence levels. 5.3.3. Conditioning on country characteristics Because we are estimating a cross-section, we obviously are precluded from using panel estimators, such as country fixed and random effects, to control for differences in country characteristics outside of our theory that may independently influence total factor productivity growth. To account for these other possible influences, we introduce a number of conditioning variables into our specification from the Sachs and Warner (1997) data set.24 The conditioning variables introduced are Sub-Sahara, a dummy indicating Sub-Saharan African nations, Landlocked, a dummy indicating a nation lacking navigable access to the sea, Tropics, a variable measuring the share of land area subject to
23 Average stocks are estimated using simple averages of period beginning and ending values. 24 See Sachs and Warner (1997) for original data sources.
Ch. 13:
Human Capital and Technology Diffusion
957
Table 3 Regression results: log H 1960–1995 Model 1
Model 2
Model 3
Model 4
C
−0.0030 (0.0024)
–
−0.0030 (0.0024)
–
ln(H 1960–1995 )
0.0175∗∗ (0.0046)
0.0150∗∗ (0.0039)
0.0184∗∗ (0.0026)
0.0159∗∗ (0.0017)
−0.0129∗∗ (0.0039)
−0.0116∗∗ (0.0036)
−0.0135∗∗ (0.0031)
−0.0122∗∗ (0.0029)
1.151 (0.783)
1.192 (0.862)
1
1
84
84
84
84
log likelihood
274.5
273.7
274.4
273.6
Wald P -value
0.00
0.00
0.00
0.00
TFP s ln(H 1960–1995 ) · TFPmi
s
Number of observations
Note: Estimation by maximum likelihood with standard errors is presented in parentheses. ∗∗ Statistical significance at the 5% confidence level.
a tropical climate, Life, the log of life expectancy at birth measured between 1965 and 1970, Ethling, a measure of ethnolinguistic fractionalization, and Openness, an indicator of the degree to which domestic policy favors free trade. We first present our results with all of the conditioning variables included, and then sequentially drop the Sub-Sahara and Openness variables. Our results are shown in Table 4. Note that the inclusion of these conditioning variables reduces our sample size from 84 to 75 countries. Models 1 and 2 report our results for our base specifications with all of the conditioning variables included. It can be seen that human capital in log levels is not positive at a statistically significant level in either specification. This result is attributable more to a substantial increase in our standard error estimate rather than a change in the point estimate of the coefficient, which does not change much in value. On the other hand, it appears that the catch-up term is robust to the inclusion of these conditioning variables, as it enters significantly with a negative coefficient at a 5 percent confidence level, as predicted. Finally, our point estimates of s are still close to 1. We cannot reject that s is negative at standard confidence levels when our intercept term is included, but we can with it excluded (Model 2).25
25 To determine whether the differences here were attributable to the inclusion of the conditioning variables or
the reduction in sample size, we estimated our models with the smaller 75 country sample reported here with
958
J. Benhabib and M.M. Spiegel Table 4 Regression results: log H 1960–1995 and geo-political variables
C ln(H 1960–1995 ) ln(H 1960–1995 ) TFP s × TFPmi s Sub-Sahara Access Tropics Life1 Ethling Openness
Number of observations log likelihood Wald P -value
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
−0.0671 (0.0463) 0.0077 (0.0061) −0.0196∗∗ (0.0066) 0.9302 (0.5796) −0.0041 (0.0030) −0.0018 (0.0027) −0.0070∗∗ (0.0026) 0.0201∗ (0.0117) 0.0001 (0.0000) 0.0128∗∗ (0.0027)
–
– 0.0067∗ (0.0038) −0.0159∗∗ (0.0041) 1.2250∗ (0.6414) –
−0.0026 (0.0026) −0.0073∗∗ (0.0026) 0.0228∗ (0.0117) 0.0000 (0.0000) 0.0128∗∗ (0.0027)
−0.0038 (0.0026) −0.0078∗∗ (0.0026) 0.0031∗∗ (0.0007) −0.0000 (0.0000) 0.0143∗∗ (0.0026)
−0.1394∗∗ (0.0488) 0.0092 (0.0077) −0.0213∗∗ (0.0082) 0.8375 (0.5857) −0.0047 (0.0032) 0.0002 (0.0029) −0.0086∗∗ (0.0027) 0.0393∗∗ (0.0123) 0.0001∗ (0.0000) –
–
0.0070 (0.0043) −0.0164∗∗ (0.0045) 1.1380∗ (0.6534) −0.0049∗ (0.0030) −0.0027 (0.0026) −0.0074∗∗ (0.0026) 0.0031∗∗ (0.0007) 0.0000 (0.0000) 0.0140∗∗ (0.0026)
−0.0778∗ (0.0461) 0.0072 (0.0054) −0.0194∗∗ (0.0059) 0.9866∗ (0.5621) –
0.0080∗∗ (0.0038) −0.0142∗∗ (0.0043) 1.2780 (0.7953) −0.0065∗∗ (0.0033) −0.0015 (0.0030) −0.0096∗∗ (0.0028) 0.0044∗∗ (0.0008) 0.0000 (0.0000) –
75 259.8 0.00
75 258.7 0.00
75 258.8 0.00
75 257.4 0.00
78 261.2 0.00
78 257.3 0.00
Note: Estimation by maximum likelihood with standard errors is presented in parentheses. ∗∗ Statistical significance at the 5% confidence level. ∗ Statistical significance at the 10% confidence level. See text for definitions of the conditioning variables.
Models 3 and 4 omit the Sub-Sahara dummy. It can be seen that human capital in log levels is still insignificant when the constant term is included, but is now significant at a 10 percent confidence level when the constant term is excluded. The catch-up term is still significantly negative at a 5 percent confidence level, as predicted. Our point estimates for s are still close to 1, with s entering significantly with a greater than zero coefficient at a 10 percent confidence level with and without the inclusion of a constant term. Finally, Models 5 and 6 omit the Openness variable. Human capital in log levels is insignificant with the constant term included, but is positive and significant, as predicted, the conditioning variables excluded. We obtained similar results to those in the larger sample. In particular, we obtained a positive and significant coefficient on human capital in log levels. This indicates that the differences in results reported here are attributable to the inclusion of the conditioning variables.
Ch. 13:
Human Capital and Technology Diffusion
959
with the exclusion of the constant term at a 10 percent confidence level. The catch-up term is still significantly negative at a 5 percent confidence level, as predicted. Our point estimates for s are again close to 1, although s is insignificantly different from zero both with and without the inclusion of a constant term in our specification. In summary, it appears that the catch-up term is strongly robust to the inclusion of the conditioning variables, while the estimates of s are still close to one, but of mixed significance. It would therefore be fair to characterize these coefficient estimates to be fairly robust to the inclusion of the conditioning variables.26 However, human capital in log levels was somewhat less robust. This result may not be surprising for a number of reasons: First, the conditioning variables, such as initial life expectancy and subsequent openness, are likely to be correlated with initial human capital levels. Indeed, initial life expectancy may be considered to be an alternative indicator of investment in human capital for many developing countries. Second, Benhabib and Spiegel (1994) found that initial human capital, which determines the rate of own-country innovation, was unimportant for a sub-sample of poorer developing countries. The introduction of our conditioning may have exposed the relatively weak role that innovation plays in total factor productivity growth for the poorer nations in our sample.27
6. Model prediction 6.1. Model forecasting Given a nation’s initial values of Hi60 and Bi (60), our transition Equation (4.4) gives us a predicted value of B at the end of our sample in 1995. Figure 2 displays the predicted values of Bi (95) conditional on Hi60 and Bi (60). One can see the logistic s form, consistent with a logistic model of technology diffusion, of our predicted values from our 26 To investigate the possibility that technological catch-up was facilitated by other variables than human cap-
ital, we substituted our Life and Openness conditioning variables for human capital in our base specification. The estimate for s was positive, but insignificant in all specifications. The coefficients on Life, both on their own and interacted with backwardness, were consistent with the theory and significant with the constant term included, but insignificant with it excluded. The coefficients on Openness, however, both on their own and interacted with backwardness, were very insignificant. As a whole, this exercise provided weak evidence of robustness for the logistic specification. Yet the imprecision of our measurements and the high correlation between country characteristic measures makes it difficult to evaluate the precise contribution of human capital relative to other potential institutional characteristics that can facilitate catch-up. For example, the correlation coefficient between hi60 and Life is 0.85. These results are available from the authors on request. 27 We also examined the robustness of our results to splitting the sample with the conditioning variables included. We split the sample into OECD and non-OECD nations. Our coefficient values for both sub-samples were of the correct sign and significant. However, we also found that the point estimate of the innovation term was larger for the OECD sub-sample, while that for the catch-up term was larger in absolute value for the nonOECD sample. This supports our findings in Benhabib and Spiegel (1994) that innovation is more important for the developed countries, while catch-up is more important for the developing nations. These regression results are also available upon request from the authors.
960
J. Benhabib and M.M. Spiegel
Figure 2. Predicted values of Bi (1995). Predicted values of Bi (95) are based on initial backwardness in TFP, Bi (60), and the log of initial stock of human capital. Bi (t) represents the ratio of TFP in country i to TFP in the leader country (United States) at time t. The sample encompasses the entire range of values for backwardness and human capital.
estimation above. Countries which have both low initial total factor productivity relative to the leader and low levels of human capital are in the low-growth portion of the plane: their predicted 1995 total factor productivity levels relative to the leader lie close to, or even below, their 1960 values. There is then a rapid acceleration in the middle range, tapering off as nations approach the total factor productivity levels of the leader. We show both the actual realizations and the predictions of our model in Figure 3. Expected values of Bi (95) for the nations in our sample based on Equation (4.4) are plotted against their realized values in 1995. The model does a fairly good job of predicting relative future productivity levels. As a measure of our goodness of fit, we calculated the coefficient of determination of the model. The ratio of residual sum-of-squared errors to the variation in the sample was only 0.115, which would correspond to an R-squared of 88.5 percent. However, there does appear to be some systematic errors in our forecasts. In particular, we seem to be systematically overestimating relative total factor productivity growth for the least backward, highest initial productivity countries like the Asian Tigers, so that the residuals for these countries are nearly all negative. This result, which suggests an even more pronounced s curve, is puzzling, but appears to leave room for future refinements in our theory. 6.2. Negative catch-up countries A more qualitative metric of the quality of fit of our model is how well it makes the discrete prediction of whether countries will be on a positive catch-up path or not. The
Ch. 13:
Human Capital and Technology Diffusion
961
Figure 3. Predicted and actual values of Bi (1995). Predicted values of Bi (95) are based on initial backwardness in TFP, Bi (60), and the log of initial stock of human capital. Bi (t) represents the ratio of TFP in country i to TFP in the leader country (United States) at time t. The sample includes observed data points only.
theory above suggests that below a certain threshold level of human capital, relative to the leader nation, a country could find its total factor productivity growth sufficiently slow that it would not exhibit convergence in total factor productivity, but would instead fall farther and farther behind the leader nation over time. In particular, we can re-write the “catch-up condition” in Equation (4.7) as sghmt Hit∗ = exp (6.1) sg + c where hmt represents the log of human capital in the leader nation at time t. Countries that find themselves with human capital stocks below Hit∗ will experience total factor productivity growth at a slower pace than the leader country. Table 5 shows the point estimates for g, c, and s based on our estimation results for Models 1 through 4 in Tables 2 and 3. As we discussed above, we cannot calculate a critical human capital stock for Model 1 in Table 2 because of our negative point estimate for g. Consequently, we concentrate on the point estimates obtained in Model 2 of Table 2, where the specification excludes a constant term independent of human capital. As we show below, our estimates of the critical human capital stocks are similar for all of our models. With the United States as our leader in total factor productivity, the point estimates obtained with Model 2 indicate that countries with average schooling in the population over the age of 25 below 1.78 years will display slower total factor productivity growth than the leader nation. We note that the critical human capital stocks were relatively insensitive to model specification or the use of initial or average human capital levels.
962
J. Benhabib and M.M. Spiegel Table 5 Point estimates Model 1
Model 2
Model 3
Model 4
g
−0.0006
0.0031
0.0012
0.0063
c
0.0198
0.0268
0.0089
0.0072
s
2.3040
3.1645
1
1
∗ H60
n.a.
1.78
1.29
2.75
∗ H95
n.a.
1.95
1.35
3.22
g
0.0046
0.0034
0.0049
0.0037
c
0.0149
0.0138
0.0135
0.0122
s
1.1515
1.1921
1
1
∗ H60
1.76
1.63
1.78
1.65
∗ H95
1.93
1.76
1.95
1.79
H1960
H 1960–1995
∗ and H ∗ represent the Note: g, c, and s are obtained from the point estimates presented in Tables 2 and 3. H60 95 minimal initial estimated stock of human capital needed for positive predicted growth relative to the leader nation.
Similarly, we can also calculate the average years of schooling in the population needed to experience faster total factor productivity growth than the United States in 1995. Because of the increase in average years of schooling in the United States, the ∗ are uniformly larger than those for H ∗ . Again using our point point estimates for Hi95 i60 estimates from Model 2, we estimate the critical level of average years of schooling in the population to be 1.95. This increase in the threshold level of human capital is due to the fact that with a larger stock of human capital, the leader nation will be innovating at a faster pace. Consequently, other nations will need to exhibit a faster pace of catch-up to experience faster total factor productivity growth than the leader. We use these estimated critical human capital stocks to conduct 2 explorations in the data. First, we can identify nations in our sample that would be predicted to exhibit slower growth in total factor productivity than the United States in 1960. This would include all nations with human capital levels in 1960 below 1.78 years of schooling. Our results are shown in Table 2. Based on our point estimates, we identify 27 nations as being below the critical human capital stock level in 1965. These nations are listed in Table 6, along with their average initial human capital stock levels.
Ch. 13:
Human Capital and Technology Diffusion
963
Table 6 Nations with slow TFP growth (1960) Country Nepal Mali Niger Mozambique Togo Central African Republic Iran Pakistan Ghana Bangladesh Algeria Syria Uganda Indonesia Papua New Guinea Kenya Cameroon Jordan Guatemala India Botswana Zimbabwe Senegal Zambia Honduras Malawi El Salvador
H1960
(TFP growthi ) – (TFP growthUSA )
0.07 0.17 0.20 0.26 0.32 0.39 0.63 0.63 0.69 0.79 0.97 0.99 1.10 1.11 1.13 1.20 1.37 1.40 1.43 1.45 1.46 1.54 1.60 1.60 1.69 1.70 1.70
−0.0072 −0.0199 −0.0297 −0.0307 −0.0172 −0.0295 −0.0050 0.0063 −0.0067 −0.0084 −0.0067 0.0034 −0.0090 0.0095 −0.0057 −0.0061 −0.0169 −0.0082 −0.0067 0.0013 0.0168 −0.0002 −0.0198 −0.0238 −0.0122 −0.0070 −0.0116
Note: The nations listed are those with 1960 human capital levels below 1.78, the minimum needed for TFP catchup according to Model 2 in Table 2.
The second column examines the growth performance of these nations over the subsequent 35 years in our sample. While it is not a formal test of our model, it is rather striking that 22 of the 27 nations predicted to exhibit slower total factor productivity growth than the United States actually did so over the course of our sample. This is markedly different than the overall sample share, where 49 of the 84 countries exhibited faster total factor productivity growth than the United States. Consequently, the subsequent performance of these nations appears to support the possibility of a logistic form of technology diffusion. Our second data exploration concerns the question of whether there are any nations that are still below the critical human capital stock, so that they are expected to have slower total factor productivity growth than the United States in the future. We investigate this question using our 1995 data. As mentioned above, the critical human capital
964
J. Benhabib and M.M. Spiegel Table 7 Nations with slow TFP growth (1995) Country Mali Niger Mozambique Nepal
H1995
TFP1995i TFP1995 USA
0.69 0.69 1.01 1.53
0.1188 0.1152 0.1499 0.1489
Note: the nations listed are those with 1995 human capital levels below 1.95, the minimum needed for TFP catchup according to Model 2 in Table 2. For the full 84 country TFP
sample, TFP 1995i = 0.4377. 1995 USA
stock using any model specification is estimated to have increased slightly between 1960 and 1995, from 1.78 average years of schooling to 1.95. Nevertheless, the good news is that because many developing nations have made substantial efforts to increase primary education rates in their populations, there are few countries who failed to meet this criterion in 1995. The four nations that fell below the critical human capital level in 1995 are listed in Table 7. They are Mali, Niger, Mozambique and Nepal. While the success of the rest of the world in acquiring sufficient human capital to be on positive catch-up path in total factor productivity is reassuring, the situation faced by these four nations is still alarming. As shown in Table 7, none of these nations has a total factor productivity level exceeding 15 percent of that in the United States. In contrast, the average ratio of the total factor productivity of a nation in our sample to that of the United States is approximately 44 percent. Our model therefore predicts that these nations will remain notably poor in the absence some sort of policy intervention.
7. Conclusion This paper generalizes the Nelson–Phelps catch-up model of technology diffusion facilitated by levels of human capital. We allow for the possibility that the pattern of technology diffusion is exponential. This specification predicts that nations will exhibit positive catch-up in growth rates. In contrast a logistic diffusion specification implies that a country with a sufficiently small capital stock may exhibit slower total factor productivity growth than the leader nation.We then derive a nonlinear specification for total factor productivity growth that nests these two specifications. We test this specification for a cross-section of 84 countries. Our results favor the logistic specification over the exponential, and other estimated parameters are consistent with our theoretical predictions. The catch-up term in our specification is robust to a number of sensitivity checks, including the use of average rather than initial levels of human capital and the inclusion of a variety of geo-political conditioning variables commonly used in the literature. This
Ch. 13:
Human Capital and Technology Diffusion
965
supports the notion that human capital plays a positive role in the determination of total factor productivity growth rates through its influence on the rate of catch-up. However, the direct performance of the human capital term on its own is somewhat less robust. Using the coefficient estimates from our parametric estimation, we then calculate the critical human capital stocks needed to achieve positive total factor productivity growth in 1960 and 1995. Our results identify 27 nations as falling below the critical human capital level in 1960, while only 4 nations remain below the critical human capital level in 1995. The historic experiences of these nations support our theory well. 22 of the 27 nations predicted to have slower growth than the leader nation (the United States) actually did so over the subsequent 35 years. This contrasts markedly with the overall experience of the nations in our sample, where 49 of the 84 nations experienced faster total factor productivity growth than the leader nation.
References Acemoglu, D. (2003). “Factor prices and technical change: From induce innovations to recent debates”. In: Aghion, P., et al. (Eds.), Knowledge, Information, and Expectations in Modern Macroeconomics: In Honor of Edmund Phelps. Princeton University Press, pp. 464–491. Acemoglu, D., Aghion, P., Zilibotti, F. (2002). “Distance to frontier, selection and economic growth”. NBER Working Paper No. w9066. Aghion, P., Christopher, H., Vickers, J. (1997). “Competition and growth with step-by-step innovation: An example”. European Economic Review 41 (3–5), 771–782. Banks, R.B. (1994). Growth and Diffusion Phenomena. Springer, Berlin. Barro, R.J., Lee, J.W. (1993). “International comparisons of educational attainment”. Journal of Monetary Economics 32, 363–394. Barro, , R.J., Sala-i-Martin, , X. (1995). Economic Growth. McGraw-Hill, Boston, MA. Barro, R.J., Sala-i-Martin, X. (1997). “Technological diffusion, convergence and growth”. Journal of Economic Growth 1, 1–26. Bartel, A.P., Lichtenberg, F.R. (1987). “The comparative advantage of educated workers in implementing new technology”. Review of Economics and Statistics 69 (1), 1–11. Basu, S., Weil, D.N. (1998). “Appropriate technology and growth”. Quarterly Journal of Economics 113 (4), 1025–1054. Benhabib, J., Spiegel, M.M. (1994). “The role of human capital in economic development: Evidence from aggregate cross-country data”. Journal of Monetary Economics 34, 143–173. Benhabib, J., Spiegel, M.M. (2000). “The role of financial development in growth and investment”. Journal of Economic Growth 5, 341–360. Bils, M., Klenow, P.J. (2000). “Does schooling cause growth?”. American Economic Review 90 (5), 1160– 1183. Branstetter, L.G. (2001). “Are knowledge spillovers international or intranational in scope? Microeconometric evidence from the U.S. Japan”. Journal of International Economics 53 (1), 53–79. Caselli, F., Esquivel, G., Lefort, F. (1996). “Reopening the convergence debate: A new look at cross-country growth empirics”. Journal of Economic Growth 1, 363–390. Coe, D.T., Helpman, E. (1995). “International R&D spillovers”. European Economic Review 39 (5), 859–887. Durlauf, S.N., Johnson, P.A. (1995). “Multiple regimes and cross-country growth behavior”. Journal of Applied Econometrics 365–384. Duffy, J., Papageorgiou, C. (2000). “A cross-country empirical investigation of the aggregate production function specification”. Journal of Economic Growth 5, 87–120.
966
J. Benhabib and M.M. Spiegel
Eaton, J., Kortum, S. (1996a). “Trade in ideas: Patenting and productivity growth in the OECD”. Journal of International Economics 40, 251–278. Eaton, J., Kortum, S. (1996b). “International technology diffusion: Theory and measurement”. International Economic Review 40, 537–570. Eeckhout, J., Jovanovic, B. (2000). “Knowledge spillovers and inequality”. American Economic Review 92 (5). Easterly, W., Loayza, N., Montiel, P. (1997). “Has Latin America’s post-reform growth been disappointing?”. Journal of International Economics 43, 287–311. Foster, A.D., Rosenzweig, M.R. (1995). “Learning by doing and learning from others: Human capital and technical change in agriculture”. Journal of Political Economy 103 (6), 1176–1209. Geroski, P.A. (1999). “Models of Technology Diffusion”. Discussion Paper 2146, Center for Economic Policy Research, London. Gerschenkron, A. (1962). Economic Backwardness in Historical Perspective. Cambridge, Belknap Press of Harvard University Press. Gollin, D. (2002). “Getting income shares right”. Journal of Political Economy 110 (2), 458–474. Griliches, Z. (1992). “The search for R&D spillovers”. Scandinavian Journal of Economics 94, 29–47. Grossman, G.M., Helpman, E. (1991). “Trade, knowledge spillovers and growth”. European Economic Review 35, 517–526. Hall, R.E., Jones, C.I. (1999). “Why do some countries produce so much more output per worker than others?”. Quarterly Journal of Economics 114 (1), 83–116. Hanushek, E.A., Kimko, D.D. (2000). “Schooling, labor-force quality, and the growth of nations”. American Economic Review 90 (5), 1184–1208. Helpman, E. (1993). “Innovation, imitation, and intellectual property rights”. Econometrica 61, 1247–1280. Howitt, P., Mayer-Foulkes, D. (2002). “R&D, implementation and stagnation: A Schumpeterian theory of convergence clubs”. NBER Working Paper Series No. 9104, Cambridge. Islam, N. (1995). “Growth empirics: A panel data approach”. Quarterly Journal of Economics 110, 1127– 1170. Keller, W. (1998). “Are international R&D spillovers trade related? Analyzing spillovers among randomly matched trade partners”. European Economic Review 42, 1469–1481. Klenow, P.J., Rodriguez-Clare, A. (1997). “The neoclassical revival in growth economics: Has it gone too far?”. NBER Macroeconomics Annual, 73–103. Krueger, A.B., Lindahl, M. (2001). “Education for growth: Why and for whom?”. Journal of Economic Literature 39 (4), 1101–1136. Mansfield, E. (1968). Industrial Research and Technological Innovation. Norton, New York. Nadiri, M.I., Kim, S. (1996). “International R&D spillovers, trade and productivity in major OECD countries”. NBER Working Paper No. 5801, October. Nelson, R.R., Phelps, E.S. (1966). “Investment in humans, technological diffusion, and economic growth”. American Economic Review 56, 69–75. Parente, S.L., Prescott, E.C. (1994). “Barriers to technology adoption and development”. Journal of Political Economy 102, 298–321. Richards, F.J. (1959). “A flexible growth function for empirical use”. Journal of Experimental Botany, 290– 300. Romer, P. (1990). “Endogenous technical change”. Journal of Political Economy 98, S71–S102. Sachs, J.D., Warner, A.M. (1997). “Fundamental sources of long-run growth”. American Economic Review 87 (2), 184–188. Segerstrom, P. (1991). “Innovation, imitation, and economic growth”. Journal of Political Economy 94, 1163– 1190. Sharif, R. (1981). “Binomial innovation diffusion models with dynamic potential adopter population”. Technological Forecasting and Social Change 20, 63–87. Temple, J. (1998). “The new growth evidence”. Journal of Economic Literature 37 (1), 112–156. Welch, F. (1975). “Human capital theory: Education, discrimination, and life cycles”. American Economic Review 65 (2), 63–73.
Chapter 14
GROWTH STRATEGIES DANI RODRIK John F. Kennedy School of Government, Harvard University, 79 Kennedy Street, Cambridge, MA 02138, USA e-mail:
[email protected]
Contents Abstract Keywords 1. Introduction 2. What we know that (possibly) ain’t so 3. The indeterminate mapping from economic principles to institutional arrangements 4. Back to the real world 4.1. In practice, growth spurts are associated with a narrow range of policy reforms 4.2. The policy reforms that are associated with these growth transitions typically combine elements of orthodoxy with unorthodox institutional practices 4.3. Institutional innovations do not travel well 4.4. Sustaining growth is more difficult than igniting it, and requires more extensive institutional reform
5. A two-pronged growth strategy 5.1. An investment strategy to kick-start growth 5.1.1. Government failures 5.1.2. Market failures 5.1.3. Where to start? 5.2. An institution building strategy to sustain growth
6. Concluding remarks Acknowledgements References
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01014-2
968 968 969 973 978 989 989 993 994 996 997 998 998 999 1002 1005 1009 1010 1010
968
D. Rodrik
Abstract This is an attempt to derive broad, strategic lessons from the diverse experience with economic growth in last fifty years. The paper revolves around two key arguments. One is that neoclassical economic analysis is a lot more flexible than its practitioners in the policy domain have generally given it credit. In particular, first-order economic principles – protection of property rights, market-based competition, appropriate incentives, sound money, and so on – do not map into unique policy packages. Reformers have substantial room for creatively packaging these principles into institutional designs that are sensitive to local opportunities and constraints. Successful countries are those that have used this room wisely. The second argument is that igniting economic growth and sustaining it are somewhat different enterprises. The former generally requires a limited range of (often unconventional) reforms that need not overly tax the institutional capacity of the economy. The latter challenge is in many ways harder, as it requires constructing over the longer term a sound institutional underpinning to endow the economy with resilience to shocks and maintain productive dynamism. Ignoring the distinction between these two tasks leaves reformers saddled with impossibly ambitious, undifferentiated, and impractical policy agendas.
Keywords economic development, institutions, policy reform JEL classification: 01, 04
Ch. 14:
Growth Strategies
969
[A]s far as the LDCs are concerned, it is probably fair to say that at least a crude sort of ‘justice’ prevails in the economic policy realm. Countries that have run their economies following the policy tenets of the professionals have on the whole reaped good fruit from the effort; likewise, those that have flown in the face of these tenets have had to pay the price. Arnold C. Harberger (1985, p. 42) When you get right down to business, there aren’t too many policies that we can say with certainty deeply and positively affect growth. Arnold C. Harberger (2003, p. 215)
1. Introduction Real per-capita income in the developing world grew at an average rate of 2.3 percent per annum during the four decades between 1960 and 2000.1 This is a high growth rate by almost any standard. At this pace incomes double every 30 years, allowing each generation to enjoy a level of living standards that is twice as high as the previous generation’s. To provide some historical perspective on this performance, it is worth noting that Britain’s per-capita GDP grew at a mere 1.3 percent per annum during its period of economic supremacy in the middle of the 19th century (1820–1870) and that the United States grew at only 1.8 percent during the half century before World War I when it overtook Britain as the world’s economic leader [Maddison (2001), Table B-22, p. 265]. Moreover, with few exceptions, economic growth in the last few decades has been accompanied by significant improvements in social indicators such as literacy, infant mortality, life expectation, and the like. So on balance the recent growth record looks quite impressive. However, since the rich countries themselves grew at a very rapid clip of 2.7 percent during the period 1960–2000, few developing countries consistently managed to close the economic gap between them and the advanced nations. As Figure 1 indicates, the countries of East and Southeast Asia constitute the sole exception. Excluding China, this region experienced per-capita GDP growth of 4.4 percent over 1960–2000. Despite the Asian financial crisis of 1997–1998 (which shows as a slight dip in Figure 1), countries such as South Korea, Thailand and Malaysia ended the century with productivity levels that stood significantly closer to those enjoyed in the advanced countries. Elsewhere, the pattern of economic performance has varied greatly across different time periods. China has been a major success story since the late 1970s, experiencing a stupendous growth rate of 8.0 percent (as compared to 2.0 percent in 1960–1980). Less spectacularly, India has roughly doubled its growth rate since the early 1980s, pulling 1 This figure refers to the exponential growth rate of GDP per capita (in constant 1995 US$) for the group of low- and middle-income countries. The data come from the World Development Indicators 2002 CD-ROM of the World Bank.
970
Figure 1. GDP per capita by country groupings (1995 US$). D. Rodrik
Ch. 14:
Growth Strategies
971
South Asia’s growth rate up to 3.3 percent in 1980–2000 from 1.2 percent in 1960–1980. The experience in other parts of the world was the mirror image of these Asian growth take-offs. Latin America and Sub-Saharan Africa both experienced robust economic growth prior to the late 1970s and early 1980s – 2.9 percent and 2.3 percent respectively – but then lost ground subsequently in dramatic fashion. Latin America’s growth rate collapsed in the “lost decade” of the 1980s, and has remained anemic despite some recovery in the 1990s. Africa’s economic decline, which began in the second half of the 1970s, continued throughout much of the 1990s and has been aggravated by the onset of HIV/AIDS and other public-health challenges. Measures of total factor productivity run parallel to these trends in per-capita output (see Table 1). Hence the aggregate picture hides tremendous variety in growth performance, both geographically and temporally. We have high growth countries and low growth countries; countries that have grown rapidly throughout, and countries that have experienced growth spurts for a decade or two; countries that took off around 1980 and countries whose growth collapsed around 1980. This paper is devoted to the question: what do we learn about growth strategies from this rich and diverse experience? By “growth strategies” I refer to economic policies and institutional arrangements aimed at achieving economic convergence with the living standards prevailing in advanced countries. My emphasis will be less on the relationship between specific policies and economic growth – the stock-in-trade of cross-national growth empirics – and more on developing a broad understanding of the contours of successful strategies. Hence my account harks back to an earlier generation of studies that distilled operational lessons from the observed growth experience, such as Albert Hirschman’s (1958), Alexander Gerschenkron’s (1962) or Walt Rostow’s (1965) books. This paper follows an unashamedly inductive approach in this tradition. A key theme in these works, as well as in the present paper, is that growth-promoting policies tend to be context specific. We are able to make only a limited number of generalizations on the effects on growth, say, of liberalizing the trade regime, opening up the financial system, or building more schools. The experience of the last two decades has frustrated the expectations of policy advisers who thought we had a good fix on the policies that promote growth – see the shift in mood that is reflected in the two quotes from Harberger that open this paper. And despite a voluminous literature, cross-national growth regressions ultimately do not provide us with much reliable and unambiguous evidence on such operational matters.2 An alternative approach, and the one I adopt here, is to shift our focus to a higher level of generality and to examine the broad design principles of successful growth strategies. This entails zooming away from the individual building blocks and concentrating on how they are put together. The paper revolves around two key arguments. One is that neoclassical economic analysis is a lot more flexible than its practitioners in the policy domain have generally
2 Easterly (2003) provides a good overview of these studies. See also Temple (1999), Brock and Durlauf (2001), and Rodríguez and Rodrik (2001).
972
D. Rodrik Table 1 Sources of growth by regions, 1960–2000 (percent increase) Contribution of:
Region/Period
Output
Output per worker
Physical capital
Education
Productivity
World (84) 1960–1970 1970–1980 1980–1990 1990–2000
5.1 3.9 3.5 3.3
3.5 1.9 1.8 1.9
1.2 1.1 0.8 0.9
0.3 0.5 0.3 0.3
1.9 0.3 0.8 0.8
Industrial Countries (22) 1960–1970 1970–1980 1980–1990 1990–2000
5.2 3.3 2.9 2.5
3.9 1.7 1.8 1.5
1.3 0.9 0.7 0.8
0.3 0.5 0.2 0.2
2.2 0.3 0.9 0.5
2.8 5.3 9.2 10.1
0.9 2.8 6.8 8.8
0.0 1.6 2.1 3.2
0.3 0.4 0.4 0.3
0.5 0.7 4.2 5.1
East Asia less China (7) 1960–1970 1970–1980 1980–1990 1990–2000
6.4 7.6 7.2 5.7
3.7 4.3 4.4 3.4
1.7 2.7 2.4 2.3
0.4 0.6 0.6 0.5
1.5 0.9 1.3 0.5
Latin America (22) 1960–1970 1970–1980 1980–1990 1990–2000
5.5 6.0 1.1 3.3
2.8 2.7 −1.8 0.9
0.8 1.2 0.0 0.2
0.3 0.3 0.5 0.3
1.6 1.1 −2.3 0.4
South Asia (4) 1960–1970 1970–1980 1980–1990 1990–2000
4.2 3.0 5.8 5.3
2.2 0.7 3.7 2.8
1.2 0.6 1.0 1.2
0.3 0.3 0.4 0.4
0.7 −0.2 2.2 1.2
Africa (19) 1960–1970 1970–1980 1980–1990 1990–2000
5.2 3.6 1.7 2.3
2.8 1.0 −1.1 −0.2
0.7 1.3 −0.1 −0.1
0.2 0.1 0.4 0.4
1.9 −0.3 −1.4 −0.5
Middle East (9) 1960–1970 1970–1980 1980–1990 1990–2000
6.4 4.4 4.0 3.6
4.5 1.9 1.1 0.8
1.5 2.1 0.6 0.3
0.3 0.5 0.5 0.5
2.6 −0.6 0.1 0.0
China (1) 1960–1970 1970–1980 1980–1990 1990–2000
Source: Bosworth and Collins (2003).
Ch. 14:
Growth Strategies
973
given it credit. In particular, first-order economic principles – protection of property rights, contract enforcement, market-based competition, appropriate incentives, sound money, debt sustainability – do not map into unique policy packages. Good institutions are those that deliver these first-order principles effectively. There is no unique correspondence between the functions that good institutions perform and the form that such institutions take. Reformers have substantial room for creatively packaging these principles into institutional designs that are sensitive to local constraints and take advantage of local opportunities. Successful countries are those that have used this room wisely. The second argument is that igniting economic growth and sustaining it are somewhat different enterprises. The former generally requires a limited range of (often unconventional) reforms that need not overly tax the institutional capacity of the economy. The latter challenge is in many ways harder, as it requires constructing a sound institutional underpinning to maintain productive dynamism and endow the economy with resilience to shocks over the longer term. Ignoring the distinction between these two tasks leaves reformers saddled with impossibly ambitious, undifferentiated, and impractical policy agendas. The plan for the paper is as follows. The next section sets the stage by evaluating the standard recipes for economic growth in light of recent economic performance. Section 3 develops the argument that sound economic principles do not map into unique institutional arrangements and reform strategies. Section 4 re-interprets recent growth experience using the conceptual framework of the previous section. Section 5 discusses a two-pronged growth strategy that differentiates between the challenges of igniting growth and the challenges of sustaining it. Concluding remarks are presented in Section 6.
2. What we know that (possibly) ain’t so Development policy has always been subject to fads and fashions. During the 1950s and 1960s, “big push”, planning, and import-substitution were the rallying cries of economic reformers in poor nations. These ideas lost ground during the 1970s to more marketoriented views that emphasized the role of the price system and outward-orientation.3 By the late 1980s a remarkable convergence of views had developed around a set of policy principles that John Williamson (1990) infelicitously termed the “Washington Consensus”. These principles remain at the heart of today’s conventional understanding of a desirable policy framework for economic growth, even though they have been greatly embellished and expanded in the years since. The left panel in Table 2 shows Williamson’s original list, which focused on fiscal discipline, “competitive” currencies, trade and financial liberalization, privatization and
3 Easterly (2001) provides an insightful and entertaining account of the evolution of thinking on economic development. See also Lindauer and Pritchett (2002) and Krueger (1997).
974
D. Rodrik Table 2 Rules of good behavior for promoting economic growth
Original Washington Consensus:
“Augmented” Washington Consensus: . . . the previous 10 items, plus:
1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
Fiscal discipline Reorientation of public expenditures Tax reform Interest rate liberalization Unified and competitive exchange rates Trade liberalization Openness to DFI Privatization Deregulation Secure property rights
Corporate governance Anti-corruption Flexible labor markets Adherence to WTO disciplines Adherence to international financial codes and standards “Prudent” capital-account opening Non-intermediate exchange rate regimes Independent central banks/inflation targeting Social safety nets Targeted poverty reduction
deregulation. These were perceived to be the key elements of what Krugman (1995, p. 29) has called the “Victorian virtue in economic policy”, namely “free markets and sound money”. Towards the end of the 1990s, this list was augmented in the thinking of multilateral agencies and policy economists with a series of so-called second-generation reforms that were more institutional in nature and targeted at problems of “good governance”. A complete inventory of these Washington Consensus-plus reforms would take too much space, and in any case the precise listing differs from source to source.4 I have shown a representative sample of ten items (to preserve the symmetry with the original Washington Consensus) in the right panel of Table 2. They range from anti-corruption and corporate governance to social safety nets and targeted anti-poverty programs. The perceived need for second-generation reforms arose from a combination of sources. First, there was growing recognition that market-oriented policies may be inadequate without more serious institutional transformation, in areas ranging from the bureaucracy to labor markets. For example, trade liberalization may not reallocate an economy’s resources appropriately if the labor markets are “rigid” or insufficiently “flexible”. Second, there was a concern that financial liberalization may lead to crises and excessive volatility in the absence of a more carefully delineated macroeconomic framework and improved prudential regulation. Hence the focus on non-intermediate exchange-rate regimes, central bank independence, and adherence to international financial codes and standards. Finally, in response to the complaint that the Washington Consensus represented a trickle-down approach to poverty, the policy framework was augmented with social policies and anti-poverty programs. It is probably fair to say that a listing along the lines of Table 2 captures in broad brushstrokes mainstream thinking about the key elements of a growth program circa 4 For diverse perspectives on what the list should contain, see Stiglitz (1998), World Bank (1998), Naim (1999), Birdsall and de la Torre (2001), Kaufmann (2002), Ocampo (2002), and Kuczynski and Williamson (2003).
Ch. 14:
Growth Strategies
975
2000. How does such a list fare when held against the light of contemporary growth experience? Imagine that we gave Table 2 to an intelligent Martian and asked him to match the growth record displayed in Figure 1 and Table 1 with the expectations that the list generates. How successful would he be in identifying which of the regions adopted the standard policy agenda and which did not? Consider first the high performing East Asian countries. Since this region is the only one that has done consistently well since the early 1960s, the Martian would reasonably guess that there is a high degree of correspondence between its policies and the list in Table 2. But he would be at best half-right. South Korea’s and Taiwan’s growth policies, to take two important illustrations, exhibit significant departures from the Washington Consensus. Neither country undertook significant deregulation or liberalization of their trade and financial systems well into the 1980s. Far from privatizing, they both relied heavily on public enterprises. South Korea did not even welcome direct foreign investment. And both countries deployed an extensive set of industrial policies that took the form of directed credit, trade protection, export subsidization, tax incentives, and other non-uniform interventions. Using the minimal scorecard of the original Washington Consensus (left panel of Table 2), the Martian would award South Korea a grade of 5 (out of 10) and Taiwan perhaps a 6 [Rodrik (1996b)]. The gap between the East Asian “model” and the more demanding institutional requirements shown on the right panel of Table 2 is, if anything, even larger. I provide a schematic comparison between the standard “ideal” and the East Asian reality in Table 3 for a number of different institutional domains such as corporate governance, financial markets, business–government relationships, and public ownership. Looking at this, the Martian might well conclude that South Korea, Taiwan, and (before them) Japan stood little chance to develop. Indeed, such were the East Asian anomalies that when the Asian financial crisis of 1997–1998 struck, many observers attributed the crisis to the moral hazard, “cronyism”, and other problems created by East Asian-style institutions [see MacLean (1999), Frankel (2000)]. The Martian would also be led astray by China’s boom since the late 1970s and by India’s less phenomenal, but still significant growth pickup since the early 1980s. While both of these countries have transformed their attitudes towards markets and private enterprise during this period, their policy frameworks bear very little resemblance to what is described in Table 2. India deregulated its policy regime slowly and undertook very little privatization. Its trade regime remained heavily restricted late into the 1990s. China did not even adopt a private property rights regime and it merely appended a market system to the scaffolding of a planned economy (as discussed further below). It is hardly an exaggeration to say that had the Chinese economy stagnated in the last couple of decades, the Martian would be in a better position to rationalize it using the policy guidance provided in Table 2 than he is to explain China’s actual performance.5
5 Vietnam, a less well known case than China, has many of the same characteristics: rapid growth since the late 1980s as a result of heterodox reform. Vietnam has benefited from a gradual turn toward markets and
976
D. Rodrik Table 3 East Asian anomalies
Institutional domain
Standard ideal
“East Asian” pattern
Property rights
Private, enforced by the rule of law
Private, but government authority occasionally overrides the law (esp. in Korea).
Corporate governance
Shareholder (“outsider”) control, protection of shareholder rights
Insider control
Business–government relations
Arms’ length, rule based
Close interactions
Industrial organization
Decentralized, competitive markets, Horizontal and vertical integration in with tough anti-trust enforcement production (chaebol); government-mandated “cartels”
Financial system
Deregulated, securities based, with free entry. Prudential supervision through regulatory oversight.
Bank based, restricted entry, heavily controlled by government, directed lending, weak formal regulation
Labor markets
Decentralized, de-institutionalized, “flexible” labor markets
Lifetime employment in core enterprises (Japan)
International capital flows
“Prudently” free
Restricted (until the 1990s)
Public ownership
None in productive sectors
Plenty in upstream industries
The Martian would be puzzled that the region that made the most determined attempt at remaking itself in the image of Table 2, namely Latin America, has reaped so little growth benefit out of it. Countries such as Mexico, Argentina, Brazil, Colombia, Bolivia, and Peru did more liberalization, deregulation and privatization in the course of a few years than East Asian countries have done in four decades. Figure 2 shows an index of structural reform for these and other Latin American countries, taken from Lora (2001a). The index measures on a scale from 0 to 1 the extent of trade and financial liberalization, tax reform, privatization, and labor-market reform undertaken. The regional average for the index rises steadily from 0.34 in 1985 to 0.58 in 1999. Yet the striking fact from Figure 1 is that Latin America’s growth rate has remained significantly below its pre-1980 level. The Martian would be at a loss to explain why growth is now lower given that the quality of Latin America’s policies, as judged by the list in Table 2,
greater reliance on private entrepreneurship, but as Van Arkadie and Mallon (2003) argue, it is hard to square the extensive role of the state and the nature of the property rights regime with the tenets of the Washington Consensus.
Ch. 14: Growth Strategies
Figure 2. Structural reform index for Latin American countries. Source: Lora (2001a). 977
978
D. Rodrik
has improved so much.6 A similar puzzle, perhaps of a smaller magnitude, arises with respect to Africa, where economic decline persists despite an overall (if less marked) “improvement” in the policy environment.7 The Martian would recognize that the growth record is consistent with some of the higher-order economic principles that inspire the standard policy consensus. A semblance of property rights, sound money, fiscal solvency, market-oriented incentives – these are elements that are common to all successful growth strategies.8 Where they have been lacking, economic performance has been lackluster at best. But the Martian would also have to conclude that the mapping from our more detailed policy preferences (such as those in Table 2) to economic success is quite imperfect. He would wonder if we cannot do better.
3. The indeterminate mapping from economic principles to institutional arrangements Here is another thought experiment. Imagine a Western economist was invited to Beijing in 1978 in order to advise the Chinese leadership on a reform strategy. What would she recommend and why? The economist would recognize that reform must start in the rural areas since the vast majority of the poor live there. An immediate recommendation would be the liberalization of agricultural markets and the abolition of the state order system under which peasants had to make obligatory deliveries of crops at low, state-controlled prices. But since price liberalization alone would be inadequate to generate the appropriate supply
6 Lora (2001b) finds that structural reforms captured by this index do correlate with growth rates in the
predicted manner, but that the impacts (taking the decade of the 1990s as a whole) are not that strong. Another econometric study by Loayza, Fajnzylber and Calderon (2002) claims that Latin America’s reforms added significantly to the region’s growth. However the latter paper uses outcome variables such as trade/GDP and financial depth ratios as its indicators of “policy”, and therefore is unable to link economic performance directly to the reforms themselves. Lin and Liu (2003) attribute the failure of the Washington Consensus to the non-viability of enterprises created under the previous “distorted” policy regime and the political impossibility of letting these go bust. 7 See also Milanovic (2003) for a closely related Martian thought experiment. Milanovic emphasizes that economic growth has declined in most countries despite greater globalization. 8 Here is how Larry Summers (2003) summarizes the recent growth evidence: “[The] rate at which countries grow is substantially determined by three things: their ability to integrate with the global economy through trade and investment; their capacity to maintain sustainable government finances and sound money; and their ability to put in place an institutional environment in which contracts can be enforced and property rights can be established. I would challenge anyone to identify a country that has done all three of these things and has not grown at a substantial rate.” Note how these recommendations are couched not in terms of specific policies (maintain tariffs below x percent, raise the government primary surplus above y percent, privatize state enterprises, and so on), but in terms of “abilities” and “capacities” to get certain outcomes accomplished. I will suggest below that these “abilities” and “capacities” do not map neatly into the standard policy preferences, and can be generated in a variety of ways.
Ch. 14:
Growth Strategies
979
Table 4 The logic of the Washington Consensus and a Chinese counterfactual Problem Low agricultural productivity Production incentives Loss of fiscal revenues Urban wages Monopoly Enterprise restructuring Unemployment . . . and so on
Solution −−−−−→ −−−−−→ −−−−−→ −−−−−→ −−−−−→ −−−−−→ −−−−−→
Price liberalization Land privatization Tax reform Corporatization Trade liberalization Financial sector reform Social safety nets
incentives under a system of communal land ownership, the economist would also recommend the privatization of land. Next, the economist would have to turn her attention to the broader implications of price liberalization in agriculture. Without access to cheap grains, the state would be left without a source of implicit tax revenue, so tax reform must be on the agenda as well. And in view of the rise of food prices, there must be a way to respond to urban workers’ demand for higher wages. State enterprises in urban areas must be corporatized, so that their managers are in a position to adjust their wages and prices appropriately. But now there are other problems that need attention. In an essentially closed and non-competitive economy, price-setting autonomy for the state behemoths entails the exercise of monopoly power. So the economist would likely recommend trade liberalization in order to “import” price discipline from abroad. Openness to trade in turn calls for other complementary reforms. There must be financial sector reform so that financial intermediaries are able to assist domestic enterprises in the inevitable adjustments that are called forth. And of course there must be social safety nets in place so that those workers who are temporarily displaced have some income support during the transition. The story can be embellished by adding other required reforms, but the message ought to be clear. By the time the Western economist is done, the reform agenda she has formulated looks very similar to the Washington Consensus (see Table 4). The economist’s reasoning is utterly plausible, which underscores the point that the Washington Consensus is far from silly: it is the result of systematic thinking about the multiple, often complementary reforms needed to establish property rights, put market incentives to work, and maintain macroeconomic stability. But while this particular reform program represents a logically consistent way achieving these end goals, it is not the only one that has the potential of doing so. In fact, in view of the administrative and political constraints that such an ambitious agenda is likely to encounter, it is not implausible that there would be better ways of getting there. How can we be sure of this? We know this because China took a very different approach to reform – one that was experimental in nature and relied on a series of institutional innovations that departed significantly from Western norms. What is important
980
D. Rodrik
to realize about these innovations is that in the end they delivered – for a period of a couple of decades at least – the very same goals that the Western economist would have been hoping for: market-oriented incentives, property rights, macroeconomic stability. But they did so in a peculiar fashion that, given the Chinese historical and political context, had numerous advantages. For example, the Chinese authorities liberalized agriculture only at the margin while keeping the plan system intact. Farmers were allowed to sell surplus crops freely at a market-determined price only after they had fulfilled their obligations to the state under the state order system. As Lau, Qian and Roland (2000) explain, this was an ingenious system that generated efficiency without creating any losers. In particular, it was a shortcut that neatly solved a conundrum inherent in wholesale liberalization: how to provide microeconomic incentives to producers while insulating the central government from the fiscal consequences of liberalization. As long as state quotas were set below the fully liberalized market outcome (so that transactions were conducted at market prices at the margin) and were not ratcheted up (so that producers did not have to worry about the quotas creeping up as a result of marketed surplus), China’s dual-track reform in effect achieved full allocative efficiency. But it entailed a different infra-marginal distribution – one that preserved the income streams of initial claimants. The dual track approach was eventually employed in other areas as well, such as industrial goods (e.g. coal and steel) and labor markets (employment contracts). Lau, Qian and Roland (2000) argue that the system was critical to achieve political support for the reform process, maintain its momentum, and minimize adverse social implications. Another important illustration comes from the area of property rights. Rather than privatize land and industrial assets, the Chinese government implemented novel institutional arrangements such as the Household Responsibility System (under which land was “assigned” to individual households according to their size) and Township and Village Enterprises (TVEs). The TVEs were the growth engine of China until the mid1990s [Qian (2003)], with their share in industrial value added rising to more than 50 percent by the early 1990s [Lin, Cai and Li (1996, p. 180)], so they deserve special comment. Formal ownership rights in TVEs were vested not in private hands or in the central government, but in local communities (townships or villages). Local governments were keen to ensure the prosperity of these enterprises as their equity stake generated revenues directly for them. Qian (2003) argues that in the environment characteristic of China, property rights were effectively more secure under direct local government ownership than they would have been under a private property-rights legal regime. The efficiency loss incurred due to the absence of private control rights was probably outweighed by the implicit security guaranteed by local government control. It is difficult to explain otherwise the remarkable boom in investment and entrepreneurship generated by such enterprises. Qian (2003) discusses other examples of “transitional institutions” China employed to fuel economic growth – fiscal contracts between central and local governments, anonymous banking – and one may expand his list by including arrangements such as Special Economic Zones. The main points to take from this experience are the
Ch. 14:
Growth Strategies
981
following. First, China relied on highly unusual, non-standard institutions. Second, these unorthodox institutions worked precisely because they produced orthodox results, namely market-oriented incentives, property rights, macroeconomic stability, and so on. Third, it is hard to argue, in view of China’s stupendous growth, that a more standard, “best-practice” set of institutional arrangements would have necessarily done better. The Chinese experience helps lay out the issues clearly because its institutional innovations and growth performance are both so stark. But China’s experience with non-standard growth policies is hardly unusual; in fact it is more the rule than the exception. The (other) East Asian anomalies noted previously (Table 3) can be viewed as part of the same pattern: non-standard practices in the service of sound economic principles. I summarize a few non-Chinese illustrations in Table 5. Consider for example the case of financial controls. I noted earlier that few of the successful East Asian countries undertook much financial liberalization early on in their development process. Interest rates remained controlled below market-clearing levels and competitive entry (by domestic or foreign financial intermediaries) was typically blocked. It is easy to construct arguments as to why this was beneficial from an economic standpoint. Table 5 summarizes the story laid out by Hellmann, Murdock and Stiglitz (1997), who coin the term “financial restraint” for the Asian model. Where asymmetric information prevails and the level of savings is sub-optimal, Hellman et al. argue that creating a moderate amount of rents for incumbent banks can generate useful incentives. These rents induce banks to do a better job of monitoring their borrowers (since there is more at stake) and to expand effort to mobilize deposits (since there are rents to be earned on them). The quality and level of financial intermediation can both be higher than under financial liberalization. These beneficial effects are more likely to materialize when the pre-existing institutional landscape has certain properties – for example when the state is not “captured” by private interests and the external capital account is restricted (see last two columns of Table 5). When these preconditions are in place, the economic logic behind financial restraint is compelling. The second illustration in Table 5 comes from South Korea’s and Taiwan’s experiences with industrial policy. The governments in these countries rejected the standard advice that they take an arm’s length approach to their enterprises and actively sought to coordinate private investments in targeted sectors. Once again, it is easy to come up with economic models that provide justification for this approach. In Rodrik (1995), I argued that the joint presence of scale economies and inter-industry linkages can depress the private return to investment in non-traditional activities below the social return. Industrial policy can be viewed as a “coordination device” to stimulate socially profitable investments. In particular, the socialization of investment risk through implicit bailout guarantees may be economically beneficial despite the obvious moral hazard risk it poses. However, once again, there are certain prerequisites and institutional complements that have to be in place for this approach to make sense (see Table 5). The third illustration in Table 5 refers to Japan and concerns the internal organization of the workplace, drawing on Aoki’s (1997) work. Aoki describes the peculiar institutional foundations of Japan’s postwar success as having evolved from a set of
982
D. Rodrik
Table 5 How to understand/rationalize institutional anomalies: four illustrations Objective
What is problem?
Institutional response
Prerequisites
Institutional complements
Financial deepening (saving mobilization and efficient intermediation)
Asymmetric information (investors know more about their projects than lenders do) and limited liability
“Financial restraint” [Hellmann, Murdock and Stiglitz (1997)]: • Controlled deposit rates and restricted entry; • Creation of rents to induce better portfolio risk management, better monitoring of firms, and increased deposit mobilization by banks.
Ability to maintain restraint at moderate levels; Positive real interest rates; Macroeconomic stability; Avoid state capture by financial interests.
Finance: Highly regulated financial markets (absence of security markets and closed capital accounts to prevent cherry picking and rent dissipation); Politics: State “autonomy” to prevent capture and decay into “crony capitalism”.
Spurring investment and entrepreneurship in non-traditional activities
Economies of scale together with inter-industry linkages depress private return to entrepreneurship/investment below social return
“Industrial policy as a A high level of coordination device” human capital relative [Rodrik (1995)]: to physical capital; • credit subsidies A relatively compet(Korea) and tax ent bureaucracy incentives (Taiwan) to select investment for selected sectors; projects. • protection of home market coupled with export subsidies; • public enterprise creation for upstream products; • arm-twisting and cajoling by political leadership; • socialization of investment risk through implicit investment guarantees.
Trade: Need to combine import protection (in selected sectors) with exposure to competition in export markets to distinguish highproductivity firms from lowproductivity ones; Business– government relations: “Embedded autonomy” (Evans) to enable close interactions and information exchange while preventing state capture and decay into “crony capitalism”.
Ch. 14:
Growth Strategies
983 Table 5 (Continued)
Objective
What is problem?
Productive organization of the workplace
Reduce anti-export bias
Institutional response
Prerequisites
Institutional complements
Trade-off “Horizontal hierarchy” between [Aoki (1997)] information sharing (working together) and economies of specialization (specialized tasks)
(Unintended) fit with prewar arrangements of military resource mobilization in Japan.
Corporate governance: insider control to provide incentive for accumulating long-term managerial skills; Labor markets: lifetime employment and enterprise unionism to generate long-term collaborative teamwork; Financial markets: main bank system to discipline firms and reduce the moral hazard consequences of insider control; Politics: “bureaupluralism” (regulation, protection) to redistribute benefits to less productive, traditional sectors.
Export processing zone Import[Rodrik (1999a, 1999b)] competing interests are politically powerful and opposed to trade liberalization
Saving boom; elastic supply of foreign investment; preferential market access in EU.
Dual labor markets: segmentation between male and female labor force, so that increase female employment in the EPZ does not drive wages up in the rest of the economy.
984
D. Rodrik
arrangements originally designed for wartime mobilization and centralized control of resources. He presents Japan’s team-centered approach to work organization and its redistribution of economic resources from advanced to backward sectors – arrangements that he terms “horizontal hierarchy” and “bureau-pluralism”, respectively – as solutions to particular informational and distributive dilemmas the Japanese economy faced in the aftermath of World War II. Unlike the previous authors, however, he views this fit between institutions and economic challenges as having been unintended and serendipitous. Lest the reader think this is solely an East Asian phenomenon, an interesting example of institutional innovation comes from Mauritius [Rodrik (1999a)]. Mauritius owes a large part of its success to the creation in 1970 of an export-processing zone (EPZ), which enabled an export boom in garments to European markets. Yet, instead of liberalizing its trade regime across the board, Mauritius combined this EPZ with a domestic sector that was highly protected until the mid-1980s, a legacy of the policies of import-substituting industrialization (ISI) followed during the 1960s. The industrialist class that had been created with these policies was naturally opposed to the opening up of the trade regime. The EPZ scheme provided a neat way around this difficulty [Wellisz and Saw (1993)]. The creation of the EPZ generated new profit opportunities, without taking protection away from the import-substituting groups. The segmentation of labor markets was particularly crucial in this regard, as it prevented the expansion of the EPZ (which employed mainly female labor) from driving wages up in the rest of the economy, and thereby disadvantaging import-substituting industries. New profit opportunities were created at the margin, while leaving old opportunities undisturbed. At a conceptual level, the story here is essentially very similar to the two-track reforms in China described earlier. To produce the results it did, however, the EPZ also needed a source of investible funds, export-oriented expertise, and market access abroad, which were in turn provided by a terms-of-trade boom, entrepreneurs from Hong Kong, and preferential market access in Europe, respectively [Rodrik (1999a); Subramanian and Roy (2003)]. In reviewing cases such as these, there is always the danger of reading too much into them after the fact. In particular, we need to avoid several fallacies. First, we cannot simply assume that institutions take the form that they do because of the functions that they perform (the functionalist fallacy). Aoki’s account of Japan is a particularly useful reminder that a good fit between form and function might be the unintended consequence of historical forces. Second, it is not correct to ascribe the positive outcomes in the cases just reviewed only to their anomalies (the ex-post rationalization fallacy). Many accounts of East Asian success emphasize the standard elements – fiscal conservatism, investment in human resources, and export orientation [see for example World Bank (1993)]. As I will discuss below, East Asian institutional anomalies have often produced perverse results when employed in other settings. And it is surely not the case that all anomalies are economically functional. The main point I take from these illustrations is robust to these fallacies, and has to do with the “plasticity” of the institutional structure that neoclassical economics is capa-
Ch. 14:
Growth Strategies
985
ble of supporting. All of the above institutional anomalies are compatible with, and can be understood in terms of, neoclassical economic reasoning (“good economics”). Neoclassical economic analysis does not determine the form that institutional arrangements should or do take. What China’s case and other examples discussed above demonstrate is that the higher-order principles of sound economic management do not map into unique institutional arrangements. In fact, principles such as appropriate incentives, property rights, sound money, and fiscal solvency all come institution-free. We need to operationalize them through a set of policy actions. The experiences above show us that there may be multiple ways of packing these principles into institutional arrangements. Different packages have different costs and benefits depending on prevailing political constraints, levels of administrative competence, and market failures. The pre-existing institutional landscape will typically offer both constraints and opportunities, requiring creative shortcuts or bold experiments. From this perspective, the “art” of reform consists of selecting appropriately from a potentially infinite menu of institutional designs. A direct corollary of this line of argument is that there is only a weak correspondence between the higher-order principles of neoclassical economics and the specific policy recommendations in the standard list (as enumerated in Table 2). To see this, consider for example one of the least contentious recommendations in the list, having to do with trade liberalization. Can the statement “trade liberalization is good for economic performance” be derived from first principles of neoclassical economics? Yes, but only if a number of side conditions are met: • The liberalization must be complete or else the reduction in import restrictions must take into account the potentially quite complicated structure of substitutability and complementarity across restricted commodities.9 • There must be no microeconomic market imperfections other than the trade restrictions in question, or if there are some, the second-best interactions that are entailed must not be adverse.10 • The home economy must be “small” in world markets, or else the liberalization must not put the economy on the wrong side of the “optimum tariff”.11 • The economy must be in reasonably full employment, or if not, the monetary and fiscal authorities must have effective tools of demand management at their disposal.
9 There is a large theoretical literature on partial trade reform, which shows the difficulty of obtaining unambiguous characterizations of the welfare effects of incomplete liberalization. See Hatta (1977), Anderson and Neary (1992), and Lopez and Panagariya (1992). For an applied general equilibrium analysis of how these issues can complicate trade reform in practice, see Harrison, Rutherford and Tarr (1993). 10 For an interesting empirical illustration on how trade liberalization can interact adversely with environmental externalities, see Lopez (1997). 11 This is not a theoretical curiosity. Gilbert and Varangis (2003) argue that the liberalization of cocoa exports in West African countries has depressed world cocoa prices, with most of the benefits being captured by consumers in developed countries.
986
D. Rodrik
• The income redistributive effects of the liberalization should not be judged undesirable by society at large, or if they are, there must be compensatory tax-transfer schemes with low enough excess burden.12 • There must be no adverse effects on the fiscal balance, or if there are, there must be alternative and expedient ways of making up for the lost fiscal revenues. • The liberalization must be politically sustainable and hence credible so that economic agents do not fear or anticipate a reversal.13 All these theoretical complications could be sidestepped if there were convincing evidence that in practice trade liberalization systematically produces improved economic performance. But even for this relatively uncontroversial policy, it has proved difficult to generate unambiguous evidence [see Rodríguez and Rodrik (2001), Vamvakidis (2002), and Yanikkaya (2003)].14 The point is that even the simplest of policy recommendations – “liberalize foreign trade” – is contingent on a large number of judgment calls about the economic and political context in which it is to be implemented.15 Such judgment calls are often made implicitly. Rendering them explicit has a double advantage: it warns us about the potential minefields that await the standard recommendations, and it stimulates creative thinking on alternatives (as in China) that can sidestep those minefields. By contrast, when the policy recommendation is made unconditionally, as in the Washington Consensus, the gamble is that the policy’s prerequisites will coincide with our actual draw from a potentially large universe of possible states of the world. I summarize this discussion with the help of Tables 6, 7, and 8 dealing with microeconomic policy, macroeconomic policy, and social policy, respectively. Each table contains three columns. The first column displays the ultimate goal that is targeted by the policies and institutional arrangements in the three domains. Hence microeconomic policies aim to achieve static and dynamic efficiency in the allocation of resources. Macroeconomic policies aim for macroeconomic and financial stability. Social policies target poverty reduction and social protection. The next column displays some of the key higher-order principles that economic analysis brings to the table. Allocative efficiency requires property rights, the rule of 12 The standard workhorse model of international trade, the factor-endowments model and its associated
Stolper–Samuelson theorem, comes with sharp predictions on the distributional effects of import liberalization (the “magnification effect”). 13 Calvo (1989) was the first to point out that lack of credibility acts as an intertemporal distortion. See also Rodrik (1991). 14 Recent empirical studies have begun to look for non-linear effects of trade liberalization. In a study of India’s liberalization, Aghion et al. (2003) find that trade liberalization appears to have generated differentiated effects across Indian firms depending on prevailing industrial capabilities and labor market regulations. Firms that were close to the technological frontier and in states with more “flexible” regulations responded positively while others responded negatively. See also Helleiner (1994) for a useful collection of country studies that underscores the contingent nature of economies’ response to trade liberalization. 15 This is one reason why policy discussions on standard recommendations such as trade liberalization and privatization now often take the formulaic form: “policy x is not a panacea; in order to work, it must be supported by reforms in the areas of a, b, c, d, and so on.”
Ch. 14:
Growth Strategies
987
Table 6 Sound economics and institutional counterparts: microeconomics Objective
Universal principles
Plausible diversity in institutional arrangements
Productive efficiency (static and dynamic)
Property rights: Ensure potential and What type of property rights? Private, current investors can retain the returns public, cooperative? to their investments. What type of legal regime? Common Incentives: Align producer incentives law? Civil law? Adopt or innovate? with social costs and benefits. What is the right balance between Rule of law: Provide a transparent, decentralized market competition and stable and predictable set of rules. public intervention? Which types of financial institutions/corporate governance are most appropriate for mobilizing domestic savings? Is there a public role to stimulate technology absorption and generation (e.g. IPR “protection”)? Table 7 Sound economics and institutional counterparts: macroeconomics
Objective
Universal principles
Plausible diversity in institutional arrangements
Macroeconomic and financial stability
Sound money: Do not generate liquidity How independent should the central bank beyond the increase in nominal money be? demand at reasonable inflation. What is the appropriate exchange-rate Fiscal sustainability: Ensure public debt regime (dollarization, currency board, remains “reasonable” and stable in adjustable peg, controlled float, pure relation to national aggregates. float)? Prudential regulation: Prevent financial Should fiscal policy be rule-bound, and if system from taking excessive risk. so what are the appropriate rules? Size of the public economy. What is the appropriate regulatory apparatus for the financial system? What is the appropriate regulatory treatment of capital account transactions?
988
D. Rodrik Table 8 Sound economics and institutional counterparts: social policy
Objective
Universal principles
Plausible diversity in institutional arrangements
Distributive justice and poverty alleviation
Targeting: Redistributive programs How progressive should the tax system should be targeted as closely as possible be? to the intended beneficiaries. Should pension systems be public or Incentive compatibility: Redistributive private? programs should minimize incentive distortions. What are the appropriate points of intervention: Educational system? Access to health? Access to credit? Labor markets? Tax system? What is the role of “social funds”? Redistribution of endowments (land reform, endowments-at-birth)? Organization of labor markets: decentralized or institutionalized? Modes of service delivery: NGOs, participatory arrangements, etc.
law, and appropriate incentives. Macroeconomic and financial stability requires sound money, fiscal solvency, and prudential regulation. Social inclusion requires incentive compatibility and appropriate targeting. These are the “universal principles” of sound economic management. They are universal in the sense that it is hard to see what any country would gain by systematically defying them. Countries that have adhered to these principles – no matter how unorthodox their manner of doing so may have been – have done well while countries that have flouted them have typically done poorly. From the standpoint of policy makers, the trouble is that these universal principles are not operational as stated. In effect, the answers to the real questions that preoccupy policy makers – how far should I go in opening up my economy to foreign competition, should I free up interest rates, should I rely on payroll taxes or the VAT, and the others listed in the third column of each table – cannot be directly deduced from these principles. This opens up space for a multiplicity of institutional arrangements that are compatible with the universal, higher-order principles. These tables clarify why the standard recommendations (Table 2) correlate poorly with economic performance around the world. The Washington Consensus, in its various forms, has tended to blur the line that separates column 2 from column 3. Policy advisors have been too quick in jumping from the higher-order principles in column 2
Ch. 14:
Growth Strategies
989
to taking unconditional stands on the specific operational questions posed in column 3. And as their policy advice has yielded disappointing results, they have moved on to recommendations with even greater institutional specificity (as with “second generation reforms”). As a result, sound economics has often been delivered in unsound form. I emphasize that this argument is not one about the advantages of gradualism over shock therapy. In fact, the set of ideas I have presented are largely orthogonal to the long-standing debate between the adherents of the two camps [see for example Lipton and Sachs (1990), Aslund, Boone and Johnson (1996), Williamson and Zagha (2002)]. The strategy of gradualism presumes that policy makers have a fairly good idea of the institutional arrangements that they want to acquire ultimately, but that for political and other reasons they can proceed only step-by-step in that direction. The argument here is that there is typically a large amount of uncertainty about what those institutional arrangements are, and therefore that the process that is required is more one of “search and discovery” than one of gradualism. The two strategies may coincide when policy changes reveal information and small-scale policy reforms have a more favorable ratio of information revelation to risk of failure.16 But it is best not to confuse the two strategies. What stands out in the real success cases, as I will further illustrate below, is not gradualism per se but an unconventional mix of standard and non-standard policies well attuned to the reality on the ground.
4. Back to the real world Previously we had asked our Martian to interpret economic performance in the real world from the lens of the standard reform agenda. Suppose we now remove the constraint and ask him to summarize the stylized facts as he sees them. Here is a list of four stylized facts that he may come up with. 4.1. In practice, growth spurts are associated with a narrow range of policy reforms One of the most encouraging aspects of the comparative evidence on economic growth is that it often takes very little to get growth started. To appreciate the point, it is enough to turn to Table 9, which lists 83 cases of growth accelerations. The table shows all cases of significant growth accelerations since the mid-1950s that can be identified statistically. The definition of a growth acceleration is the following: an increase in an economy’s per-capita GDP growth of 2 percentage points or more (relative to the previous 5 years) that is sustained over at least 8 years. The timing of the growth acceleration is determined by fitting a spline centered on the candidate break years, and selecting the
16 For example, Dewatripont and Roland (1995) and Wei (1997) present models in which gradual reforms
reveal information and affect subsequent political constraints.
990
D. Rodrik
break that maximizes the fit of the equation [see Hausmann, Pritchett and Rodrik (2004) for details on the procedure].17 Most of the usual suspects are included in the table: for example Taiwan 1961, Korea 1962, Indonesia 1967, Brazil 1967, Mauritius 1971, China 1978, Chile 1986, Uganda 1989, Argentina 1990, and so on. But the exercise also yields a large number of much less well-known cases, such as Egypt 1976 or Pakistan 1979. In fact, the large number of countries that have managed to engineer at least one instance of transition to high growth may appear as surprising. As I will discuss later, most of these growth spurts have eventually collapsed. Nonetheless, an increase in growth of 2 percent (and typically more) over the better part of a decade is nothing to sneer at, and it is worth asking what produces it. In the vast majority of the cases listed in Table 9, the “shocks” (policy or otherwise) that produced the growth spurts were apparently quite mild. Asking most development economists about the policy reforms of Pakistan in 1979 or Syria in 1969 would draw a blank stare. This reflects the fact that not much reform was actually taking pace in these cases. Relatively small changes in the background environment can yield significant increase in economic activity. Even in the well-known cases, policy changes at the outset have been typically modest. The gradual, experimental steps towards liberalization that China undertook in the late 1970s were discussed above. South Korea’s experience in the early 1960s was similar. The military government led by Park Chung Hee that took power in 1961 did not have strong views on economic reform, except that it regarded economic development as its key priority. It moved in a trial-and-error fashion, experimenting at first with various public investment projects. The hallmark reforms associated with the Korean miracle, the devaluation of the currency and the rise in interest rates, came in 1964 and fell far short of full liberalization of currency and financial markets. As these instances illustrate, an attitudinal change on the part of the top political leadership towards a more market-oriented, private-sector-friendly policy framework often plays as large a role as the scope of policy reform itself (if not larger). Perhaps the most important example of this can be found in India: such an attitudinal change appears to have had a particularly important effect in the Indian take-off of the early 1980s, which took place a full decade before the liberalization of 1991 [DeLong (2003); Rodrik and Subramanian (2004)]. This is good news because it suggests countries do not need an extensive set of institutional reforms in order to start growing. Instigating growth is a lot easier in practice than the standard recipe, with its long list of action items, would lead us to believe. This should not be surprising from a growth theory standpoint. When a country is so far below its potential steady-state level of income, even moderate movements in the right direction can produce a big growth payoff. Nothing could be more encouraging to policy makers, who are often overwhelmed and paralyzed by the apparent need to undertake policy reforms on a wide and ever-expanding front. 17 The selection strategy allows multiple accelerations, but they must be at least five years apart. We require
post-acceleration growth to be at least 3.5 percent, and also rule out recoveries from crises.
Ch. 14:
Growth Strategies
991
Table 9 Episodes of rapid growth by region, decade and magnitude of acceleration Region
Decade
Country
Year
Growth before
Growth after
Difference in growth
SubSaharan Africa
1950s and 1960s
NGA BWA GHA GNB ZWE COG NGA MUS TCD CMR COG UGA LSO RWA MLI MWI GNB MUS UGA MWI
1967 1969 1965 1969 1964 1969 1957 1971 1973 1972 1978 1977 1971 1975 1972 1970 1988 1983 1989 1992
−1.7 2.9 −0.1 −0.3 0.6 0.9 1.2 −1.8 −0.7 −0.6 3.1 −0.6 0.7 0.7 0.8 1.5 −0.7 1.0 −0.8 −0.8
7.3 11.7 8.3 8.1 7.2 5.4 4.3 6.7 7.3 5.3 8.2 4.0 5.3 4.0 3.8 3.9 5.2 5.5 3.6 4.8
9.0 8.8 8.4 8.4 6.5 4.5 3.0 8.5 8.0 5.9 5.1 4.6 4.6 3.3 3.0 2.5 5.9 4.4 4.4 5.6
PAK PAK LKA IND
1962 1979 1979 1982
−2.4 1.4 1.9 1.5
4.8 4.6 4.1 3.9
7.1 3.2 2.2 2.4
THA KOR IDN SGP TWN CHN MYS MYS THA PNG KOR IDN CHN
1957 1962 1967 1969 1961 1978 1970 1988 1986 1987 1984 1987 1990
−2.5 0.6 −0.8 4.2 3.3 1.7 3.0 1.1 3.5 0.3 4.4 3.4 4.2
5.3 6.9 5.5 8.2 7.1 6.7 5.1 5.7 8.1 4.0 8.0 5.5 8.0
7.8 6.3 6.2 4.0 3.8 5.1 2.1 4.6 4.6 3.7 3.7 2.1 3.8
1970s
1980s and 1990s
South Asia
1950s/1960s 1970s 1980s
East Asia
1950s and 1960s
1970s 1980s and 1990s
992
D. Rodrik Table 9 (Continued)
Region
Decade
Country
Year
Growth before
Growth after
Difference in growth
Latin America and Caribbean
1950s and 1960s
DOM BRA PER PAN NIC ARG COL ECU PRY TTO PAN URY CHL URY HTI ARG DOM
1969 1967 1959 1959 1960 1963 1967 1970 1974 1975 1975 1974 1986 1989 1990 1990 1992
−1.1 2.7 0.8 1.5 0.9 0.9 1.6 1.5 2.6 1.9 2.6 1.5 −1.2 1.6 −2.3 −3.1 0.4
5.5 7.8 5.2 5.4 4.8 3.6 4.0 8.4 6.2 5.4 5.3 4.0 5.5 3.8 12.7 6.1 6.3
6.6 5.1 4.4 3.9 3.8 2.7 2.4 6.8 3.7 3.5 2.7 2.6 6.7 2.1 15.0 9.2 5.8
MAR SYR TUN ISR ISR JOR EGY SYR DZA
1958 1969 1968 1967 1957 1973 1976 1974 1975
−1.1 0.3 2.1 2.8 2.2 −3.6 −1.6 2.6 2.1
7.7 5.8 6.6 7.2 5.3 9.1 4.7 4.8 4.2
8.8 5.5 4.5 4.4 3.1 12.7 6.3 2.2 2.1
SYR
1989
−2.9
4.4
7.3
ESP DNK JPN USA CAN IRL BEL NZL AUS FIN FIN PRT ESP IRL GBR FIN NOR
1959 1957 1958 1961 1962 1958 1959 1957 1961 1958 1967 1985 1984 1985 1982 1992 1991
4.4 1.8 5.8 0.9 0.6 1.0 2.1 1.5 1.5 2.7 3.4 1.1 0.1 1.6 1.1 1.0 1.4
8.0 5.3 9.0 3.9 3.6 3.7 4.5 3.8 3.8 5.0 5.6 5.4 3.8 5.0 3.5 3.7 3.7
3.5 3.5 3.2 3.0 2.9 2.7 2.4 2.4 2.3 2.2 2.2 4.3 3.7 3.4 2.5 2.8 2.2
1970s
1980s and 1990s
Middle East and North Africa
1950s and 1960s
1970s
1980s and 1990s OECD
1950s and 1960s
1980s and 1990s
Source: Hausmann, Pritchett and Rodrik (2004).
Ch. 14:
Growth Strategies
993
4.2. The policy reforms that are associated with these growth transitions typically combine elements of orthodoxy with unorthodox institutional practices No country has experienced rapid growth without minimal adherence to what I have termed higher-order principles of sound economic governance – property rights, market-oriented incentives, sound money, fiscal solvency. But as I have already argued, these principles were often implemented via policy arrangements that are quite unconventional. I illustrated this using examples such as China’s two-track reform strategy, Mauritius’ export processing zone, and South Korea’s system of “financial restraint”. It is easy to multiply the examples. When Taiwan and South Korea decided to reform their trade regimes to reduce anti-export bias, they did this not via import liberalization (which would have been a Western economist’s advice) but through selective subsidization of exports. When Singapore decided to make itself more attractive to foreign investment, it did this not by reducing state intervention but by greatly expanding public investment in the economy and through generous tax incentives [Young (1992)]. Botswana, which has an admirable record with respect to macroeconomic stability and the management of its diamond wealth, also has one of the largest levels of government spending (in relation to GDP) in the world. Chile, a country that is often cited as a paragon of virtue by the standard check list, has also departed from it in some important ways: it has kept its largest export industry (copper) under state ownership; it has maintained capital controls on financial inflows through the 1990s; and it has provided significant technological, organizational, and marketing assistance to its fledgling agro-industries. In all these instances, standard desiderata such as market liberalization and outward orientation were combined with public intervention and selectivity of some sort. The former element in the mix ensures that any economist so inclined can walk away from the success cases with a renewed sense that the standard policy recommendations really “work”. Most egregiously, China’s success is often attributed to its turn towards market – which is largely correct – and then with an unjustified leap of logic is taken as a vindication of the standard recipe – which is largely incorrect. It is not clear how helpful such evaluations are when so much of what these countries did is unconventional and fits poorly with the standard agenda.18 It is difficult to identify cases of high growth where unorthodox elements have not played a role. Hong Kong is probably the only clear-cut case. Hong Kong’s government has had a hands-off attitude towards the economy in almost all areas, the housing market being a major exception. Unlike Singapore, which followed a free trade policy but otherwise undertook extensive industrial policies, Hong Kong’s policies have been as 18 Another source of confusion is the mixing up of policies with outcomes. Successful countries end up with
much greater participation in the world economy, thriving private sectors, and a lot of financial intermediation. What we need to figure out, however, are the policies that produce these results. It would be a great distortion of the strategy followed by countries such as China, South Korea, Taiwan and others to argue that these outcomes were the result of trade and financial liberalization, and privatization.
994
D. Rodrik
close to laissez-faire as we have ever observed. However, there were important prerequisites to Hong Kong’s success, which illuminate once again the context-specificity of growth strategies. Most important, Hong Kong’s important entrepôt role in trade, the strong institutions imparted by the British, and the capital flight from communist China had already transformed the city-state into a high investment, high entrepreneurship economy by the late 1950s. As Figure 3 shows, during the early 1960s Hong Kong’s investment rate was more than three times higher than that in South Korea or Taiwan. The latter two economies would not reach Hong Kong’s 1960 per-capita GDP until the early 1970s.19 Hence Hong Kong did not face the same challenge that Taiwan, South Korea, and Singapore did to crowd in private investment and stimulate entrepreneurship. It goes without saying that not all unorthodox remedies work. And those that work sometimes do so only for a short while. Consider for example Argentina’s experiment in the 1990s with a currency board. Most economists would consider a currency board regime as too risky for an economy of Argentina’s size insofar as it prevents expenditure switching via the exchange rate. (Hong Kong has long operated a successful marketing board.) However, as the Argentinean economy began to grow rapidly in the first half of the 1990s, many analysts altered their views. Had the Asian crisis of 1997–1998 and the Brazilian devaluation of 1999 not forced Argentina off its currency board, it would have been easy to construct a story ex post about the virtues of the currency board as a growth strategy. The currency board sought to counteract the effects of more than a century of financial mismanagement through monetary discipline. It was a shortcut aimed at convincing foreign and domestic investors that the rules of the game had changed irrevocably. Under better external circumstances, the credibility gained might have more than offset the disadvantages. The problem in this case was the unwillingness to pull back from the experiment even when it became clear that the regime had left the Argentine economy with a hopelessly uncompetitive real exchange rate. The lesson is that institutional innovation requires a pragmatic approach which avoids ideological lock-in. 4.3. Institutional innovations do not travel well The more discouraging aspect of the stylized facts is that the policy packages associated with growth accelerations – and particularly the elements therein that are non-standard – tend to vary considerably from country to country. China’s two-track strategy of reform differs significantly from India’s gradualism. South Korea’s and Taiwan’s more protectionist trade strategy differs markedly from the open trade policies of Singapore (and Hong Kong). Even within strategies that look superficially similar, closer look reveals large variation. Taiwan and South Korea both subsidized non-traditional industrial activities, but the former did it largely through tax incentives and the latter largely through directed credit.20 19 These and investment data are from the Penn World Tables 6.1. 20 On the institutional differences among East Asian economies, see Haggard (2003).
Ch. 14: Growth Strategies
Figure 3. Investment as a share of GDP in East Asia. 995
996
D. Rodrik
Attempts to emulate successful policies elsewhere often fail. When Gorbachev tried to institute a system similar to China’s Household Responsibility System and two-track pricing in the Soviet Union during the mid- to late-1980s, it produced few of the beneficial results that China had obtained.21 Most developing countries have export processing zones of one kind or another, but few have been as successful as the one in Mauritius. Import-substituting industrialization (ISI) worked in Brazil, but not in Argentina.22 In light of the arguments made earlier, this experience should not be altogether surprising. Successful reforms are those that package sound economic principles around local capabilities, constraints and opportunities. Since these local circumstances vary, so do the reforms that work. An immediate implication is that growth strategies require considerable local knowledge. It does not take a whole lot of reform to stimulate economic growth – that is the good news. The bad news is that it may be quite difficult to identify where the binding constraints or promising opportunities lie. A certain amount of policy experimentation may be required in order to discover what will work. China represents the apotheosis of this experimental approach to reform. But it is worth noting that many other instances of successful reform were preceded by failed experiments. In South Korea, President Park’s developmental efforts initially focused on the creation of white elephant industrial projects that ultimately went nowhere [Soon (1994, pp. 27– 28)]. In Chile, Pinochet’s entire first decade can be viewed as a failed experiment in “global monetarism”. Economists can have a useful role to play in this process: they can identify the sources of inefficiency, describe the relevant trade offs, figure out general-equilibrium implications, predict behavioral responses, and so on. But they can do these well only if their analysis is adequately embedded within the prevailing institutional and political reality. The hard work needs to be done at home. 4.4. Sustaining growth is more difficult than igniting it, and requires more extensive institutional reform The main reason that few of the growth accelerations listed in Table 9 are etched in the consciousness of development economists is that most of them did not prove durable. In fact, as discussed earlier, over the last four decades few countries except for a few East Asian ones have steadily converged to the income levels of the rich countries. The vast majority of growth spurts tend to run out of gas after a while. The experience of Latin America since the early 1980s and the even more dramatic collapse of SubSaharan Africa are emblematic of this phenomenon. In a well-known paper, Easterly et al. (1993) were the first to draw attention to a related finding, namely the variability in 21 Murphy, Shleifer and Vishny (1992) analyze this failure and attribute it to the inability of the Soviet state
to enforce the plan quotas once market pricing was allowed (albeit at the margin). This had been critical to the success of the Chinese approach. 22 TFP growth averaged 2.9 and 0.2 percent per annum in Brazil and Argentina, respectively, during 1960– 1973. See Rodrik (1999a) and Collins and Bosworth (1996).
Ch. 14:
Growth Strategies
997
growth performance across time periods. The same point is made on a broader historical canvas by Goldstone (in preparation). Hence growth in the short- to medium-term does not guarantee success in the longterm. A plausible interpretation is that the initial reforms need to be deepened over time with efforts aimed at strengthening the institutional underpinning of market economies. It would be nice if a small number of policy changes – which, as argued above, is what produces growth accelerations – could produce growth over the longer term as well, but this is obviously unrealistic. I will discuss some of the institutional prerequisites of sustained growth in greater detail later in the paper. But the key to longer-term prosperity, once growth is launched, is to develop institutions that maintain productive dynamism and generate resilience to external shocks. For example, the growth collapses experienced by many developing countries in the period from the mid-1970s to the early 1980s seem to be related mainly to the inability to adjust to the volatility exhibited by the external environment at that time. In these countries, the effects of terms-of-trade and interest-rate shocks were magnified by weak institutions of conflict management [Rodrik (1999b)]. This, rather than the nature of microeconomic incentive regimes in place (e.g., import substituting industrialization), is what caused growth in Africa and Latin America to grind to a halt after the mid1970s and early 1980s (respectively). The required macroeconomic policy adjustments set off distributive struggles and proved difficult to undertake. Similarly, the weakness of Indonesia’s institutions explains why that country could not extricate itself from the 1997–1998 East Asian financial crisis [see Temple (2003)], while South Korea, for example, did a rapid turnaround. These examples are also a warning that continued growth in China cannot be taken for granted: without stronger institutions in areas ranging from financial markets to political governance, the Chinese economy may well find itself having outgrown its institutional underpinnings.23
5. A two-pronged growth strategy As the evidence discussed above reveals, growth accelerations are feasible with minimal institutional change. The deeper and more extensive institutional reforms needed for long-term convergence take time to implement and mature. And they may not be the most effective way to raise growth at the outset because they do not directly target the most immediate constraints and opportunities facing an economy. At the same time, such institutional reforms can be much easier to undertake in an environment of growth rather than stagnation. These considerations suggest that successful growth strategies are based on a two-pronged effort: a short-run strategy aimed at stimulating growth, and
23 Young (2000) argues that China’s reform strategy may have made things worse in the long run, by increas-
ing the number of distorted margins.
998
D. Rodrik
a medium- to long-run strategy aimed at sustaining growth.24 The rest of this section takes these up in turn. 5.1. An investment strategy to kick-start growth From the standpoint of economic growth, the most important question in the short run for an economy stuck in a low-activity equilibrium is: how do you get entrepreneurs excited about investing in the home economy? “Invest” here has to be interpreted broadly, as referring to all the activities that entrepreneurs undertake, such as expanding capacity, employing new technology, producing new products, searching for new markets, and so on. As entrepreneurs become energized, capital accumulation and technological change are likely to go hand in hand – too entangled with each other to separate out cleanly. What sets this process into motion? There are two kinds of views on this in the literature. One approach emphasizes the role of government-imposed barriers to entrepreneurship. In this view, policy biases towards large and politically-connected firms, institutional failures (in the form of licensing and other regulatory barriers, inadequate property rights and contract enforcement), and high levels of policy uncertainty and risk create dualistic economic structures and repress entrepreneurship. The removal of the most egregious forms of these impediments is then expected to unleash a flurry of new investments and entrepreneurship. According to the second view, the government has to play a more pro-active role than simply getting out of the private sector’s way: it needs to find means of crowding in investment and entrepreneurship with some positive inducements. In this view, economic growth is not the natural order of things, and establishing a fair and level playing field may not be enough to spur productive dynamism. The two views differ in the importance they attach to prevailing, irremovable market imperfections and their optimism with regard to governments’ ability to design and implement appropriate policy interventions. 5.1.1. Government failures A good example of the first view is provided by the strategy of development articulated in Stern (2001). In a deliberate evocation of Hirschman’s (1958) book, Stern outlines an approach with two pillars: building an appropriate “investment climate” and “empowering poor people”. The former is the relevant part of his approach in this context. Stern defines “investment climate” quite broadly, as “the policy, institutional, and behavioral environment, both present and expected, that influences the returns and risks associated with investment” [Stern (2001, pp. 144–145)]. At the same time, he recognizes the need
24 A similar distinction is also made by Ocampo (2003), who emphasizes that many of the long-run correlates
of growth (such as improved institutions) are the result, and not the instigator, of growth. There is also an analogue in the political science literature in the distinction between the political prerequisites of initiating and sustaining reform [see Haggard and Kaufman (1983)].
Ch. 14:
Growth Strategies
999
for priorities and the likelihood that these priorities will be context specific. He emphasizes the favorable dynamics that are unleashed once a few, small things are done right. In terms of actual policy content, Stern’s illustrations make clear that he views the most salient features of the investment climate to be government-imposed imperfections: macroeconomic instability and high inflation, high government wages that distort the functioning of labor markets, a large tax burden, arbitrary regulations, burdensome licensing requirements, corruption, and so on. The strategy he recommends is to use enterprise surveys and other techniques to uncover which of these problems bite the most, and then to focus reforms on the corresponding margin. Similar perspectives can be found in Johnson, McMillan and Woodruff (2000), Friedman et al. (2000), and Aslund and Johnson (2003). Besley and Burgess (2002a) provide evidence across Indian states on the productivity depressing effects of labor market regulations. The title of Shleifer and Vishny’s (1998) book aptly summarizes the nature of the relevant constraint in this view: The Grabbing Hand: Government Pathologies and Their Cures. 5.1.2. Market failures The second approach focuses not on government-imposed constraints, but on market imperfections inherent in low-income environments that block investment and entrepreneurship in non-traditional activities. In this view, economies can get stuck in a low-level equilibrium due to the nature of technology and markets, even when government policy does not penalize entrepreneurship. There are many versions of this latter approach, and some of the main arguments are summarized in the taxonomy presented in Table 10. I distinguish here between stories that are based on learning spillovers (a non-pecuniary externality) and those that are based on market-size externalities induced by scale economies. See also the useful discussion of these issues in Ocampo (2003), which takes a more overtly structuralist perspective. Table 10 A taxonomy of “natural” barriers to industrialization A. Learning externalities
B. Coordination failures (market-size externalities induced by IRS)
1. Learning-by-doing [e.g., Matsuyama (1992)] 1. Wage premium in manufacturing [e.g., Murphy, 2. Human capital externalities [e.g., Azariadis and Shleifer and Vishny (1989)] Drazen (1990)] 2. Infrastructure [e.g., Murphy, Shleifer and Vishny 3. Learning about costs [e.g., Hausmann and Rodrik (1989)] (2002)] 3. Specialized intermediate inputs [e.g., Rodrik (1991, 1995)] 4. Spillovers associated with wealth distribution [e.g., Hoff and Stiglitz (2001)]
1000
D. Rodrik
As Acemoglu, Aghion and Zilibotti (2002) point out, two types of learning are relevant to economic growth: (a) adaptation of existing technologies; and (b) innovation to create new technologies. Early in the development process, the kind of learning that matters the most is of the first type. There are a number of reasons why such learning can be subject to spillovers. There may be a threshold level of human capital beyond which the private return to acquiring skills becomes strongly positive [as in Azariadis and Drazen (1990)]. There may be learning-by-doing which is either external to individual firms, or cannot be properly internalized due to imperfections in the market for credit [as in Matsuyama (1992)]. Or there may be learning about a country’s own cost structure, which spills over from the incumbents to later entrants [as in Hausmann and Rodrik (2002)]. In all these cases, the relevant learning is under-produced in a decentralized equilibrium, with the consequence that the economy fails to diversify into non-traditional, more advanced lines of activity.25 There then exist policy interventions that can improve matters. With standard externalities, the first-best takes the form of a corrective subsidy targeted at the relevant distorted margin. In practice, revenue, administrative or informational constraints may make resort to second-best interventions inevitable. For example, Hausmann and Rodrik (2002) suggest a carrot-and-stick strategy to deal with the learning barrier to industrialization that they identify. In that model, costs of production in non-traditional activities are uncertain, and they are revealed only after an upfront investment by an incumbent. Once that initial investment is made, the cost information becomes public knowledge. Entrepreneurs engaged in the cost discovery process incur private costs, but provide social benefits that can vastly exceed their anticipated profits. The first-best policy here, which is an entry subsidy, suffers from an inextricable moral hazard problem. Subsidized entrants have little incentive to engage subsequently in costly activities to discover costs. A second-best approach takes the form of incentives contingent on good performance. Hausmann and Rodrik (2002) evaluate East Asian and Latin American industrial policies from this perspective. They argue that East Asian policies were superior in that they effectively combined incentives with discipline. The former was provided through subsidies and protection, while the latter was provided through government monitoring and the use of export performance as a productivity yardstick. Latin American firms under import substituting industrialization (ISI) received considerable incentives, but faced very little discipline. In the 1990s, these same firms arguably faced lots of discipline (exerted through foreign competition), but little incentives. This line of argument provides one potential clue to the disappointing economic performance of Latin America in the 1990s despite a much improved “investment climate” according to the standard criteria.
25 Imbs and Wacziarg (2003) demonstrate that sectoral diversification is a robust correlate of economic
growth at lower levels of income. This is in tension with standard models of trade and specialization under constant returns to scale. Sectoral concentration starts to increase only after a relatively high level of income is reached, with the turning point coming somewhere between $8,500 and $9,500 in 1985 U.S. dollars.
Ch. 14:
Growth Strategies
1001
The second main group of stories shown in Table 10 relates to the existence of coordination failures induced by scale economies. The big-push theory of development, articulated first by Rosenstein-Rodan (1943) and formalized by Murphy, Shleifer and Vishny (1989), is based on the idea that moving out of a low-level steady state requires coordinated and simultaneous investments in a number of different areas. A general formulation of such models can be provided as follows. Let the level of profits in a given modern-sector activity depend on n, the proportion of the economy that is already engaged in modern activities: π m (n), with dπ m (n)/dn > 0. Let profits in traditional activities be denoted π t . Suppose modern activities are unprofitable for an individual entrant if no other entrepreneur already operates in the modern sector, but highly profitable if enough entrepreneurs do so: π m (0) < π t and π m (1) > π t . Then n = 0 and n = 1 are both possible equilibria, and industrialization may never take hold in an economy that starts with n = 0. The precise mechanism that generates profit functions of this form depends on the model in question. Murphy, Shleifer and Vishny (1989) develop models in which the complementarity arises from demand spillovers across final goods produced under scale economies or from bulky infrastructure investments. RodriguezClare (1996), Rodrik (1996a, 1996b), and Trindade (2003) present models in which the effect operates through vertical industry relationships and specialized intermediate inputs. Hoff and Stiglitz (2001) discuss a large class of models with coordination failure characteristics. The policy implications of such models can be quite unconventional, requiring the crowding in of private investment through subsidization, jawboning, public enterprises and the like. Despite the “big push” appellation, the requisite policies need not be wide-ranging. For example, socializing investment risk through implicit investment guarantees, a policy followed in South Korea, is welfare enhancing in Rodrik’s (1996a) framework because it induces simultaneous entry into the modern sector. It is also costless to the government, because the guarantees are never called on insofar as the resulting investment boom pays for itself. Hence, when successful, such policies will leave little trail on government finances or elsewhere.26 Both types of models listed in Table 10 suggest that the propagation of modern, non-traditional activities is not a natural process and that it may require positive inducements. One such inducement that has often worked in the past is a sizable and sustained depreciation of the real exchange rate. For a small open economy, the real exchange rate is defined as the relative price of tradables to non-tradables. In practice, this price ratio tends to move in tandem with the nominal exchange rate, the price of foreign currency in terms of home currency. Hence currency devaluations (supported by appropriate monetary and fiscal policies) increase the profitably of tradable activities across the board. From the current perspective, this has a number of distinct advantages. Most of the gains from diversification into non-traditional activities are likely to lie within manufactures
26 On South Korea’s implicit investment guarantees, see Amsden (1989). During the Asian financial crisis,
these guarantees became an issue and they were portrayed as evidence of crony capitalism [MacLean (1999)].
1002
D. Rodrik
and natural resource based products (i.e., tradables) rather than services and other nontradables. Second, the magnitude of the inducement can be quite large, since sustained real depreciations of 50 percent or more are quite common. Third, since tradable activities face external competition, the activities that are encouraged tend to be precisely the ones that face the greatest market discipline. Fourth, the manner in which currency depreciation subsidizes tradable activities is completely market-friendly, requiring no micromanagement on the part of bureaucrats. For all these reasons, a credible, sustained real exchange rate depreciation may constitute the most effective industrial policy there is. Large real exchange rate changes have played a big role in some of the more recent growth accelerations. Figure 4 shows two well-known cases: Chile and Uganda since the mid-1980s. In both cases, a substantial swing in relative prices in favor of tradables accompanied the growth take-off. In Chile, the more than doubling of the real exchange rate following the crisis of 1982–1983 (the deepest in Latin America at the time) is commonly presumed to have played an instrumental role in promoting diversification into non-traditional exports and stimulating economic growth. It is worth noting that import tariffs were raised significantly as well (during 1982–1985), giving importsubstituting activities an additional boost. As the bottom panel of Figure 3 shows, the depreciation in Uganda was even larger. These depreciations are unlikely to have been the result of growth, since growth typically generates an appreciation of the real exchange rate through the Balassa–Samuelson effect. By contrast, large real depreciations did not play a major role in early growth accelerations in East Asia during the 1960s [Rodrik (1997)].27 5.1.3. Where to start? The two sets of views outlined above – the government failure and market failure approaches – can help frame policy discussions and identify important ways of thinking about policy priorities in the short run. The most effective point of leverage for stimulating growth obviously depends on local circumstances. It is tempting to think that the right first step is to remove government-imposed obstacles to entrepreneurial activity before worrying about “crowding in” investments through positive inducements. But this may not always be a better strategy. Certainly when inflation is in triple digits or the regulatory framework is so cumbersome that it stifles any private initiative, removing these distortions will be the most sensible initial step. But beyond that, it is difficult to say in general where the most effective margin for change lies. Asking businessmen their views on the priorities can be helpful, but not decisive. When learning spillovers and coordination failures block economic take-off, enterprise surveys are unlikely to be revealing unless the questions are very carefully crafted to elicit relevant responses.
27 Polterovich and Popov (2002) provide theory and evidence on the role of real exchange rate undervalua-
tions in generating economic growth.
Ch. 14:
Growth Strategies
1003
(a)
(b) Figure 4. Real exchange rate and per-capita GDP growth shown as 3-year moving average in (a) Chile and (b) Uganda.
Hausmann, Rodrik and Velasco (2004) outline a framework for undertaking “growth diagnostics”, i.e., targeting reforms on the most binding constraints on economic growth. One of the lessons of recent economic history is that creative interventions can be remarkably effective even when the “investment climate”, judged by standard criteria, is pretty lousy. South Korea’s early reforms took place against the background of a po-
1004
D. Rodrik
litical leadership that was initially quite hostile to the entrepreneurial class.28 China’s TVEs have been stunningly successful despite the absence of private property rights and an effective judiciary. Conversely, the Latin American experience of the 1990s indicates that the standard criteria do not guarantee an appropriate investment climate. Governments can certainly deter entrepreneurship when they try to do too much; but they can also deter entrepreneurship when they do too little. It is sometimes argued that heterodoxy requires greater institutional strength and therefore lies out of reach of most developing countries. But the evidence does not provide much support for this view. It is true that the selective interventions I have discussed in the case of South Korea and Taiwan were successful in part due to unusual and favorable circumstances. But elsewhere, heterodoxy served to make virtue out of institutional weakness. This is the case with China’s TVEs, Mauritius’ export processing zone, and India’s gradualism. In these countries, it was precisely institutional weakness that rendered the standard remedies impractical. It is in part because the standard reform agenda is institutionally so highly demanding – a fact now recognized through the addition of so-called “second generation reforms” – that successful growth strategies are so often based on unconventional elements (in their early stages at least). It is nonetheless true that the implementation of the market failure approach requires a reasonably competent and non-corrupt government. For every South Korea, there are many Zaires where policy activism is an excuse for politicians to steal and plunder. Finely-tuned policy interventions can hardly be expected to produce desirable outcomes in setting such as the latter. And to the extent that Washington Consensus policies are more conducive to honest behavior on the part of politicians, they may well be preferable on this account. However, the evidence is ambiguous on this. Most policies, including those of the Washington Consensus type, are corruptible if the underlying political economy permits or encourages it. Consider for example Russia’s experiment with mass privatization. It is widely accepted that this process was distorted and delegitimized by asset grabs on the part of politically well-connected insiders. Washington Consensus policies themselves cannot legislate powerful rent-seekers out of existence. Rank ordering different policy regimes requires a more fully specified model of political economy than the reduced-form view that automatically associates governmental restraint with less rent-seeking.29 I close this section with the usual refrain: the range of strategies that have worked in the past is quite diverse. Traditional import-substituting industrialization (ISI) model 28 One month after taking power in a military coup in 1961, President Park arrested some of the leading
businessmen in Korea under the newly passed Law for Dealing with Illicit Wealth Accumulation. These businessmen were subsequently set free under the condition that they establish new industrial firms and give up the shares to the government [Amsden (1989, p. 72)]. 29 In Rodrik (1995) I compared export subsidy regimes in six countries, and found that the regimes that were least likely to be open to rent-seeking ex ante – those with clear-cut rules, uniform schedules, and no arm’s length relationships between firms and bureaucrats – were in fact less effective ex post. Where bureaucrats were professional and well-monitored, discretion was not harmful. Where they were not, the rules did not help.
Ch. 14:
Growth Strategies
1005
was quite effective in stimulating growth in a large number of developing countries (e.g., Brazil, Mexico, Turkey). So was East Asian style outward orientation, which combined heavy-handed interventionism at home with single-minded focus on exports (South Korea, Taiwan). Chile’s post-1983 strategy was based on quite a different style of outward orientation, relying on large real depreciation, absence of explicit industrial policies (but quite a bit of support for non-traditional exports in agro-industry), saving mobilization through pension privatization, and discouragement of short-term capital inflows. The experience of countries such as China and Mauritius is best described as two-track reform. India comes as close to genuine gradualism as one can imagine. Hong Kong represents probably the only case where growth has taken place without an active policy of crowding in private investment and entrepreneurship, but here too special and favorable preconditions (mentioned earlier) limit its relevance to other settings. In view of this diversity, any statement on what ignites growth has to be cast at a sufficiently high level of generality. 5.2. An institution building strategy to sustain growth In the long run, the main thing that ensures convergence with the living standards of advanced countries is the acquisition of high-quality institutions. The growth-spurring strategies described above have to be complemented over time with a cumulative process of institution building to ensure that growth does not run out of steam and that the economy remains resilient to shocks. This point has now been amply demonstrated both by historical accounts [North and Thomas (1973), Engerman and Sokoloff (1994)] and by econometric studies [Hall and Jones (1999), Acemoglu, Johnson and Robinson (2001), Rodrik, Subramanian and Trebbi (2002), Easterly and Levine (2002)]. However, these studies tend to remain at a very aggregate level of generality and do not provide much policy guidance [a point that is also made in Besley and Burgess (2002b)]. The empirical research on national institutions has generally focused on the protection of property rights and the rule of law. But one should think of institutions along a much wider spectrum. In its broadest definition, institutions are the prevailing rules of the game in society [North (1990)]. High quality institutions are those that induce socially desirable behavior on the part of economic agents. Such institutions can be both informal (e.g., moral codes, self-enforcing agreements) and formal (legal rules enforced through third parties). It is widely recognized that the relative importance of formal institutions increases as the scope of market exchange broadens and deepens. One reason is that setting up formal institutions requires high fixed costs but low marginal costs, whereas informal institutions have high marginal costs [Li (1999); Dixit (2004, Chapter 3)]. I will focus here on formal institutions. What kind of institutions matter and why? Table 11 provides a taxonomy of marketsustaining institutions, associating each type of institutions with a particular need. The starting point is the recognition that markets need not be self-creating, self-regulating, self-stabilizing, and self-legitimizing. Hence, the very existence of market exchange presupposes property rights and some form of contract enforcement. This is the as-
1006
D. Rodrik Table 11 A taxonomy of market-sustaining institutions
Market-creating institutions
Market-regulating institutions
Market-stabilizing institutions
Market-legitimizing institutions
Property rights
Regulatory bodies
Democracy
Contract enforcement
Other mechanisms for correcting market failures
Monetary and fiscal institutions Institutions of prudential regulation and supervision
Social protection and social insurance
pect of institutions that has received the most scrutiny in empirical work. The central dilemma here is that a political entity that is strong enough to establish property rights and enforce contracts is also strong enough, by definition, to violate these same rules for its own purpose [Djankov et al. (2003)]. The relevant institutions must strike the right balance between disorder and dictatorship. As Table 11 makes clear, there are other needs as well. Every advanced economy has discovered that markets require extensive regulation to minimize abuse of market power, internalize externalities, deal with information asymmetries, establish product and safety standards, and so on. They also need monetary, fiscal, and other arrangements to deal with the business cycle and the problems of unemployment/inflation that are at the center of macroeconomists’ analyzes since Keynes. Finally, market outcomes need to be legitimized through social protection, social insurance, and democratic governance most broadly [Rodrik (2000)]. Institutional choices made in dealing with these challenges often have to strike a balance between competing objectives. The regulatory regime governing the employment relationship must trade off the gains from “flexibility” against the benefits of stability and predictability. The corporate governance regime must delineate the interests and prerogatives of shareholders and stakeholders. The financial system must be free to take risks, but not so much so that it becomes an implicit public liability. There must be enough competition to ensure static allocative efficiency, but also adequate prospect of rents to spur innovation. The last two centuries of economic history in today’s rich countries can be interpreted as an ongoing process of learning how render capitalism more productive by supplying the institutional ingredients of a self-sustaining market economy: meritocratic public bureaucracies, independent judiciaries, central banking, stabilizing fiscal policy, antitrust and regulation, financial supervision, social insurance, political democracy. Just as it is silly to think of these as the prerequisites of economic growth in poor countries, it is equally silly not to recognize that such institutions eventually become necessary to achieve full economic convergence. In this connection, one may want to place special emphasis on democratic institutions and civil liberties, not only because they are important in and of themselves, but also because they can be viewed as metainstitutions that help society make appropriate selections from the available menu of economic institutions.
Ch. 14:
Growth Strategies
1007
However, the earlier warning not to confuse institutional function and institutional form becomes once again relevant here. Appropriate regulation, social insurance, macroeconomic stability and the like can be provided through diverse institutional arrangements. While one can be sure that some types of arrangements are far worse than others, it is also the case that many well-performing arrangements are functional equivalents. Function does not map uniquely into form. It would be hard to explain otherwise how social systems that are so different in their institutional details as those of the United States, Japan, and Europe have managed to generate roughly similar levels of wealth for their citizens. All these societies protect property rights, regulate product, labor, and financial markets, have sound money, and provide for social insurance. But the rules of the game that prevail in the American style of capitalism are very different from those in the Japanese style of capitalism. Both differ from the European style. And even within Europe, there are large differences between the institutional arrangements in, say, Sweden and Germany. There has been only modest convergence among these arrangements in recent years, with the greatest amount of convergence taking place probably in financial market practices and the least in labor market institutions [Freeman (2000)]. There are a number of reasons for institutional non-convergence. First, differences in social preferences, say over the trade-off between equity and opportunity, may result in different institutional choices. If Europeans have a much greater preference for stability and equity than Americans, their labor market and welfare-state arrangements will reflect that preference. Second, complementarities among different parts of the institutional landscape can generate hysteresis and path dependence. An example of this would be the complementarity between corporate governance and financial market practices of the Japanese “model”, as discussed previously. Third, the institutional arrangements that are required to promote economic development can differ significantly, both between rich and poor countries and among poor countries. This too has been discussed previously. There is increasing recognition in the economics literature that high-quality institutions can take a multitude of forms and that economic convergence need not necessarily entail convergence in institutional forms [North (1994), Freeman (2000), Pistor (2000), Mukand and Rodrik (2005), Berkowitz, Pistor and Richard (2003), Djankov et al. (2003), Dixit (2004)].30 North (1994, p. 8) writes: “Economies that adopt the formal rules of another economy will have very different performance characteristics than the first economy because of different informal norms and enforcement [with the implication that] transferring the formal political and economic rules of successful Western economies to third-world and Eastern European economies is not a sufficient condition for good economic performance.” Freeman (2000) discusses the variety of labor market institutions that prevail among the advanced countries and argues that differences in
30 Furthermore, as Roberto Unger (1998) has argued, there is no reason to suppose that today’s advanced
economies have already exhausted all the useful institutional variations that could underpin healthy and vibrant economies.
1008
D. Rodrik
these practices have first-order distributional effects, but only second-order efficiency effects. Pistor (2000) provides a general treatment of the issue of legal transplantation, and shows how importation of laws can backfire. In related work, Berkowitz, Pistor and Richard (2003) find that countries that developed their formal legal orders internally, adapted imported codes to local conditions, or had familiarity with foreign codes ended up with much better legal institutions than those that simply transplanted formal legal orders from abroad. Djankov et al. (2003) base their discussion on an “institutional possibility frontier” that describes the trade-off between private disorder and dictatorship, and argue that different circumstances may call for different choices along this frontier. And [Dixit (2004, p. 4)] summarizes the lessons for developing countries thus: “it is not always necessary to create replicas of western style state legal institutions from scratch; it may be possible to work with such alternative institutions as are available, and build on them.” Mukand and Rodrik (2005) develop a formal model to examine the costs and benefits of institutional “experimentation” versus “copycatting” when formulas that have proved successful elsewhere may be unsuitable at home. A key idea is that institutional arrangements that prove successful in one country create both positive and negative spillovers for other countries. On the positive side, countries whose underlying conditions are sufficiently similar to those of the successful “leaders” can imitate the arrangements prevailing there and forego the costs of experimentation. This is one interpretation of the relative success that transition economies in the immediate vicinity of the European Union have experienced. Countries such as Poland, the Czech Republic or the Baltic republics share a similar historical trajectory with the rest of Europe, have previous experience with capitalist market institutions, and envisaged full EU membership within a reasonable period [De Menil (2003)]. The wholesale adoption of EU’s acquis communautaire may have been the appropriate institution-building strategy for these countries. On the other hand, countries may be tempted or forced to imitate institutional arrangements for political or other reasons, even when their underlying conditions are too dissimilar for the strategy to make sense.31 Institutional copycatting may have been useful for Poland, but it is much less clear that it was relevant or practical for Ukraine or Kyrgyzstan. The negative gradient in the economic performance of transition economies as one moves away from Western Europe provides some support for this idea [see Mukand and Rodrik (2005)]. Even though it is recent, this literature opens up a new and exciting way of looking at institutional reform. In particular, it promises an approach that is less focused on so-called best practices or the superiority of any particular model of capitalism, and more cognizant of the context-specificity of desirable institutional arrangements. Dixit’s (2004) monograph outlines a range of theoretical models that help structure our thinking along these lines. 31 In Mukand and Rodrik (2005) it is domestic politics that generates inefficient imitation. Political leaders
may want to signal their type (and increase the probability of remaining in power) by imitating standard policies even when they know these will not work as well as alternative arrangements. But one can also appeal to the role of IMF and World Bank conditionality in producing this kind of outcome.
Ch. 14:
Growth Strategies
1009
6. Concluding remarks Richard Feynman, the irreverent physicist who won the Nobel Prize in 1965 for his work on quantum electrodynamics, relates the following story. Following the award ceremony and the dinner in Stockholm, he wanders into a room where a Scandinavian princess is holding court. The princess recognizes him as one of the awardees and asks him what he got the prize for. When Feynman replies that his field is physics, the princess says that this is too bad. Since no one at the table knows anything about physics, she says, they cannot talk about it. Feynman disagrees: “On the contrary,” I answered. “It’s because somebody knows something about it that we can’t talk about physics. It’s the things that nobody knows anything about that we can discuss. We can talk about the weather; we can talk about social problems; we can talk about psychology; we can talk about international finance . . . so it’s the subject that nobody knows anything about that we can all talk about!” [Feynman (1985)] This is not the place to defend international finance (circa 1965) against the charge Feynman levels at it. But suppose Feynman had picked on economic growth instead of international finance. Would growth economists have a plausible riposte? Is the reason we all talk so much about growth that we understand so little about it? It is certainly the case that growth theory is now a much more powerful tool than it was before Solow put pencil to paper. And cross-country regressions have surely thrown out some useful correlations and stylized facts. But at least at the more practical end of things – how do we make growth happen? – things have turned out to be somewhat disappointing. By the mid-1980s, policy oriented economists had converged on a new consensus regarding the policy framework for growth. We thought we knew a lot about what governments needed to do. But as my Martian thought experiment at the beginning of the paper underscores, reality has been unkind to our expectations. If Latin America was booming today and China and India were stagnating, we would have an easier time fitting the world to our policy framework. Instead, we are straining to explain why unorthodox, two-track, gradualist reform paths have done so much better than sure-fire adoption of the standard package. Very few policy analysts think that the answer is to go back to old-style ISI, even though its record was certainly respectable for a very large number of countries. Certainly no-one believes that central planning is a credible alternative. But by the same token, few are now convinced that liberalization, deregulation, and privatization on their own hold the key to unleashing economic growth. Maybe the right approach is to give up looking for “big ideas” altogether [as argued explicitly by Lindauer and Pritchett (2002), and implicitly by Easterly (2001)]. But that would be overshooting too. Economics is full of big ideas on the importance of incentives, markets, budget constraints, and property rights. It offers powerful ways of analyzing the allocative and distributional consequences of proposed policy changes. The key is to realize that these principles do not translate directly into specific policy recommendations. That translation requires
1010
D. Rodrik
the analyst to supply many additional ingredients that are contingent on the economic and political context, and cannot be done a priori. Local conditions matter not because economic principles change from place to place, but because those principles come institution free and filling them out requires local knowledge. Therefore, the real lesson for the architects of growth strategies is to take economics more seriously, not less seriously. But the relevant economics is that of the seminar room, with its refusal to make unconditional generalizations and its careful examination of the contingent relation between the economic environment and policy implications. Rule-of-thumb economics, which has long dominated thinking on growth policies, can be safely discarded.
Acknowledgements I gratefully acknowledge financial support from the Carnegie Corporation of New York. I also thank, without implicating, Philippe Aghion, Richard Freeman, Steph Haggard, Ricardo Hausmann, Murat Iyigun, Sharun Mukand, José Antonio Ocampo, Andrei Shleifer, and Arvind Subramanian for comments that substantially improved this paper.
References Acemoglu, D., Aghion, P., Zilibotti, F. (2002). “Distance to frontier, selection, and economic growth”. NBER Working Paper No. 9066, July. Acemoglu, D., Johnson, S., Robinson, J.A. (2001). “The colonial origins of comparative development: An empirical investigation”. American Economic Review 91 (5), 1369–1401. Aghion, P., Burgess, R., Redding, S., Zilibotti, F. (2003). “The unequal effects of liberalization: Theory and evidence from India”. Department of Economics, London School of Economics, March. Amsden, A.H. (1989). Asia’s Next Giant: South Korea and Late Industrialization. Oxford University Press, New York and Oxford. Anderson, J.E., Neary, J.P. (1992). “Trade reform with quotas, partial rent retention, and tariffs”. Econometrica 60, 57–76. Aoki, M. (1997). “Unintended fit: Organizational evolution and government design of institutions in Japan”. In: Aoki, M. et al. (Eds.), The Role of Government in East Asian Economic Development: Comparative Institutional Analysis. Clarendon Press, Oxford. Aslund, A., Johnson, S. (2003). “Small enterprises and economic policy”. Working Paper. Sloan School, MIT. Aslund, A., Boone, P., Johnson, S. (1996). “How to stabilize: Lessons from post-communist countries”. Brookings Papers on Economic Activity 1. Azariadis, C., Drazen, A. (1990). “Threshold externalities in economic development”. Quarterly Journal of Economics 105, 501–526. Berkowitz, D., Pistor, K., Richard, J.-F. (2003). “Economic development, legality, and the transplant effect”. European Economic Review 47 (1), 165–195. Besley, T., Burgess, R. (2002a). “Can labor regulation hinder economic performance? Evidence from India”. CEPR Discussion Paper No. 3260. Besley, T., Burgess, R. (2002b). “Halving global poverty”. Department of Economics, London School of Economics, August.
Ch. 14:
Growth Strategies
1011
Birdsall, N., de la Torre, A. (2001). “Washington contentious: Economic policies for social equity in Latin America”. Washington: Carnegie Endowment for International Peace and Inter-American Dialogue. Bosworth, B., Collins, S.M. (2003). “The empirics of growth: An update”. Brookings Institutions, unpublished paper. Brock, W.A., Durlauf, S.N. (2001). “Growth empirics and reality”. The World Bank Economic Review 15 (2), 229–272. Calvo, G. (1989). “Incredible reforms”. In: Calvo et al. (Eds.), Debt Stabilization and Development. Basil Blackwell, New York. Collins, S., Bosworth, B. (1996). “Economic growth in East Asia: Accumulation versus assimilation”. Brookings Papers on Economic Activity 2, 135–191. DeLong, J.B. (2003). “India since independence: An analytic growth narrative”. In: Rodrik, D. (Ed.), In Search of Prosperity: Analytic Narratives of Economic Growth. Princeton University Press, Princeton, NJ. De Menil, G. (2003). “History, policy, and performance in two transition economies: Poland and Romania”. In: Rodrik, D. (Ed.), In Search of Prosperity: Analytic Narratives of Economic Growth. Princeton University Press, Princeton, NJ. Dewatripont, M., Roland, G. (1995). “The design of reform packages under uncertainty”. American Economic Review 85 (5), 1207–1223. Dixit, A. (2004). Lawlessness and Economics: Alternative Modes of Economic Governance. Gorman Lectures. Princeton University Press. Djankov, S., Glaeser, E., La Porta, R., Lopez-de-Silanes, F., Shleifer, A. (2003). “The new comparative economics”. Harvard University, January. Easterly, W. (2001). The Elusive Quest for Growth. MIT Press, Cambridge, MA. Easterly, W. (2003). “National policies and economic growth: A reappraisal”. New York University, Development Research Institute (DRI), Working Paper No. 1. Easterly, W., Levine, R. (2002). “Tropics, germs, and crops: How endowments influence economic development”. Mimeo, Center for Global Development and Institute for International Economics. Easterly, W., Kremer, M., Pritchett, L., Summers, L.H. (1993). “Good policy or good luck? Country growth performance and temporary shocks”. Journal of Monetary Economics 32 (3), 459–483. Engerman, S.L., Sokoloff, K.L. (1994). “Factor endowments, institutions, and differential paths of growth among New World economies: A view from economic historians of the United States”. NBER Working Paper No. H0066. Feynman, R.P. (1985). Surely You’re Joking Mr. Feynman!. Norton, New York. Frankel, J. (2000). “The Asian model, the miracle, the crisis, and the fund”. In: Krugman, P. (Ed.), Currency Crises. The University of Chicago Press for the NBER. Freeman, R.B. (2000). “Single peaked vs. diversified capitalism: The relation between economic institutions and outcomes”. NBER Working Paper No. W7556. Friedman, E., Johnson, S., Kaufmann, D., Zoido-Lobaton, P. (2000). “Dodging the grabbing hand: The determinants of unofficial activity in 69 countries”. Journal of Public Economics 76, 459–493. Gerschenkron, A. (1962). Economic Backwardness in Historical Perspective: A Book of Essays. Harvard University Press, Cambridge, MA. Gilbert, C.L., Varangis, P. (2003). “Globalization and international commodity trade with specific reference to West African cocoa producers”. NBER Working Paper No. w9668. Goldstone, J.A. The Happy Chance: The Rise of the West in Global Context. U.C. Davis, pp. 1500–1850. Book manuscript in preparation. Haggard, S. (2003). “Institutions and Growth in East Asia”. Unpublished manuscript, UCSD. Haggard, S., Kaufman, R. (Eds.) (1983). The Politics of Economic Adjustment. Princeton University Press, Princeton, NJ. Hall, R., Jones, C.I. (1999). “Why do some countries produce so much more output per worker than others?”. Quarterly Journal of Economics 114 (1), 83–116. Harberger, A.C. (1985). Economic Policy and Economic Growth. International Center for Economic Growth, Institute for Contemporary Studies, San Francisco, CA.
1012
D. Rodrik
Harberger, A.C. (2003). “Interview with Arnold Harberger: Sound policies can free up natural forces of growth”. In: IMF Survey. International Monetary Fund, Washington, DC, pp. 213–216. July 14. Harrison, G.W., Rutherford, T.F., Tarr, D.G. (1993). “Trade reform in the partially liberalized economy of Turkey”. The World Bank Economic Review 7 (2), 191–218. Hatta, T. (1977). “A recommendation for a better tariff structure”. Econometrica 45, 1859–1869. Hausmann, R., Rodrik, D. (2002). “Economic development as self-discovery”. NBER Discussion Paper No. w8952. Hausmann, R., Pritchett, L., Rodrik, D. (2004). “Growth accelerations”. NBER Working Paper No. w10566. Hausmann, R., Rodrik, D., Velasco, A. (2004). “Growth diagnostics”. Unpublished paper. Helleiner, G.K. (Ed.) (1994). Trade Policy and Industrialization in Turbulent Times. Routledge, London. Hellmann, T., Murdock, K., Stiglitz, J. (1997). “Financial restraint: Toward a new paradigm”. In: Aoki, M. et al. (Eds.), The Role of Government in East Asian Economic Development: Comparative Institutional Analysis. Clarendon Press, Oxford. Hirschman, A.O. (1958). The Strategy of Economic Development. Yale University Press, New Haven, CT. Hoff, K., Stiglitz, J. (2001). “Modern economic theory and development”. In: Meier, G.M., Stiglitz, J.E. (Eds.), Frontiers of Development Economics. Oxford University Press, New York, pp. 389–459. Imbs, J., Wacziarg, R. (2003). “Stages of diversification”. American Economic Review 93 (1), 63–86. Johnson, S., McMillan, J., Woodruff, C. (2000). “Entrepreneurs and the ordering of institutional reform: Poland, Slovakia, Romania, Russia, and Ukraine compared”. Economics of Transition. Kaufmann, D. (2002). “Rethinking governance”. World Bank Institute, World Bank, Washington, DC, December. Krueger, A.O. (1997). “Trade policy and development: How we learn”. The American Economic Review, March. Krugman, P. (1995). “Dutch tulips and emerging markets”. Foreign Affairs, July/August. Kuczynski, P.-P., Williamson, J. (Eds.) (2003). After the Washington Consensus: Restarting Growth and Reform in Latin America. Institute for International Economics, Washington, DC. Lau, L.J., Qian, Y., Roland, G. (2000). “Reform without losers: An interpretation of China’s dual-track approach to transition”. The Journal of Political Economy 108 (1), 120–143. Li, S. (1999). “The benefits and costs of relation-based governance: An explanation of the East Asian miracle and crisis”. City University of Hong Kong, October. Lin, J.Y., Cai, F., Li, Z. (1996). The China Miracle: Development Strategy and Economic Reform. The Chinese University Press, Shatin, NT, Hong Kong. Lin, J.Y., Liu, M. (2003). “Development strategy, viability and challenges of development in lagging regions”. Paper prepared for the 15th World Bank’s Annual Bank Conference on Development Economics, Bangalore, India, May. Lindauer, D.L., Pritchett, L. (2002). “What’s the big idea? The third generation of policies for economic growth”. Economia, 1–40. Lipton, D., Sachs, J. (1990). “Creating a market economy in Eastern Europe: The case of Poland”. Brookings Papers on Economic Activity 1. Loayza, N., Fajnzylber, P., Calderon, C. (2002). “Economic growth in Latin America and the Caribbean. Stylized facts, explanations, and forecasts”. World Bank, Washington, DC, June. Lopez, R. (1997). “Environmental externalities in traditional agriculture and the impact of trade liberalization: The case of Ghana”. Journal of Development Economics 53 (1), 17–39. Lopez, R., Panagariya, A. (1992). “On the theory of piecemeal tariff reform: The case of pure imported intermediate inputs”. American Economic Review 82 (3), 615–625. Lora, E. (2001a). “Structural reforms in Latin America: What has been reformed and how to measure it”. Inter-American Development Bank, Washington, DC, December. Lora, E. (2001b). “El crecimiento económico en América Latina después de una década de reformas estructurales”. Washington, DC, United States: Inter-American Development Bank, Research Department. Mimeographed document. MacLean, B.K. (1999). “The rise and fall of the ‘crony capitalism’ hypothesis: Causes and consequences”. Department of Economics, Laurentian University, Ontario, March.
Ch. 14:
Growth Strategies
1013
Maddison, A. (2001). The World Economy: A Millennial Perspective. OECD Development Centre, Paris. Matsuyama, K. (1992). “Agricultural productivity, comparative advantage, and economic growth”. Journal of Economic Theory, December, 317–334. Milanovic, B. (2003). “The two faces of globalization: Against globalization as we know it”. World Development 31 (4), 667–683. Mukand, S., Rodrik, D. (2005). “In search of the Holy Grail: Policy convergence, experimentation, and economic performance”. American Economic Review 95 (1), 374–383. Murphy, K.M., Shleifer, A., Vishny, R.W. (1989). “Industrialization and the big push”. Journal of Political Economy 97 (5), 1003–1026. Murphy, K.M., Shleifer, A., Vishny, R.W. (1992). “The transition to a market economy: Pitfalls of partial reform”. The Quarterly Journal of Economics 107 (3), 889–906. Naim, M. (1999). “Fads and fashion in economic reforms: Washington Consensus or Washington Confusion?”. Paper prepared for the IMF Conference on Second Generation Reforms, Washington, DC, October. North, D.C. (1990). Institutions, Institutional Change and Economic Performance. Cambridge University Press, New York. North, D.C. (1994). “Economic performance through time”. The American Economic Review 84 (3), 359– 368. North, D.C., Thomas, R. (1973). The Rise of the Western World: A New Economic History. Cambridge University Press, Cambridge. Ocampo, J.A. (2002). “Rethinking the development agenda”. United Nations Economic Commission for Latin America and the Caribbean (ECLAC), Santiago, Chile. Ocampo, J.A. (2003). “Structural dynamics and economic growth in developing countries”. United Nations Economic Commission for Latin America and the Caribbean (ECLAC), Santiago, Chile. Pistor, K. (2000). “The standardization of law and its effect on developing economies”. G-24 Discussion Paper No. 4, July. Polterovich, V., Popov, V. (2002). “Accumulation of foreign exchange reserves and long term growth”. Unpublished paper. New Economic School, Moscow, Russia. Qian, Y. (2003). “How reform worked in China”. In: Rodrik, D. (Ed.), In Search of Prosperity: Analytic Narratives of Economic Growth. Princeton University Press, Princeton, NJ. Rodríguez, F., Rodrik, D. (2001). “Trade policy and economic growth: A skeptic’s guide to the cross-national evidence”. In: Bernanke, B., Rogoff, K.S. (Eds.), Macroeconomics Annual 2000. MIT Press for NBER, Cambridge, MA. Rodriguez-Clare, A. (1996). “The division of labor and economic development”. Journal of Development Economics 49 (April), 3–32. Rodrik, D. (1991). “Policy uncertainty and private investment in developing countries”. Journal of Development Economics 36 (November). Rodrik, D. (1995). “Taking trade policy seriously: Export subsidization as a case study in policy effectiveness”. In: Deardorff, A., Levinson, J., Stern, R. (Eds.), New Directions in Trade Theory. University of Michigan Press, Ann Arbor. Rodrik, D. (1996a). “Coordination failures and government policy: A model with applications to East Asia and Eastern Europe”. Journal of International Economics 40 (1–2), 1–22. Rodrik, D. (1996b). “Understanding economic policy reform”. Journal of Economic Literature XXXIV (March), 9–41. Rodrik, D. (1997). “Trade strategy, exports, and investment: Another look at East Asia”. Pacific Economic Review, February. Rodrik, D. (1999a). The New Global Economy and Developing Countries: Making Openness Work. Overseas Development Council, Washington, DC. Rodrik, D. (1999b). “Where did all the growth go? External shocks, social conflict and growth collapses”. Journal of Economic Growth, December. Rodrik, D. (2000). “Institutions for high-quality growth: What they are and how to acquire them”. Studies in Comparative International Development 35 (3).
1014
D. Rodrik
Rodrik, D., Subramanian, A. (2004). “From ‘Hindu growth’ to productivity surge: The mystery of the Indian growth transition”. NBER Working Paper No. w10376, March. Rodrik, D., Subramanian, A., Trebbi, F. (2002). “Institutions rule: The primacy of institutions over geography and integration in economic development”. Kennedy School of Government, Harvard University (October). Rosenstein-Rodan, P. (1943). “Problems of industrialization of Eastern and Southeastern Europe”. Economic Journal 53 (210–211), 202–211. Rostow, W.W. (1965). The Stages of Economic Growth: A Non-Communist Manifesto. Cambridge University Press, Cambridge and New York. Shleifer, A., Vishny, R.W. (1998). The Grabbing Hand: Government Pathologies and Their Cures. Harvard University Press, Cambridge, MA. Soon, C. (1994). The Dynamics of Korean Development. Institute for International Economics, Washington, DC. Stern, N. (2001). “A strategy for development”. ABCDE Keynote Address, Washington, DC, World Bank (May). Stiglitz, J.E. (1998). “More instruments and broader goals moving toward the post-Washington consensus”. United Nations University/WIDER, Helsinki. Subramanian, A., Roy, D. (2003). “Who can explain the Mauritian miracle? Meade, Romer, Sachs, or Rodrik?”. In: Rodrik, D. (Ed.), In Search of Prosperity: Analytic Narratives of Economic Growth. Princeton University Press, Princeton, NJ. Summers, L.H. (2003). “Godkin lectures”. John F. Kennedy School of Government, Harvard University (April). Temple, J. (1999). “The new growth evidence”. Journal of Economic Literature 37 (1), 112–156. Temple, J. (2003). “Growing into trouble: Indonesia since 1966”. In: Rodrik, D. (Ed.), In Search of Prosperity: Analytic Narratives of Economic Growth. Princeton University Press, Princeton, NJ. Trindade, V. (2003). “The big push, industrialization, and international trade: The role of exports”. Maxwell School, Syracuse University, March. Unger, R.M. (1998). Democracy Realized: The Progressive Alternative. Verso, London and New York. Vamvakidis, A. (2002). “How robust is the growth–openness connection? Historical evidence”. Journal of Economic Growth 7 (1), 57–80. Van Arkadie, B., Mallon, R. (2003). Vietnam: A Transition Tiger?. Asia Pacific Press at The Australian National University, Australia. Wei, S.-J. (1997). “Gradualism versus Big Bang: Speed and sustainability of reforms”. Canadian Journal of Economics 30 (4B), 1234–1247. Wellisz, S., Saw, P.L.S. (1993). “Mauritius”. In: Findlay, R., Wellisz, S. (Eds.), The Political Economy of Poverty, Equity, and Growth: Five Open Economies. Oxford University Press, New York. Williamson, J. (1990). “What Washington means by policy reform”. In: Williamson, J. (Ed.), Latin American Adjustment: How Much Has Happened? Institute for International Economics, Washington. Williamson, J., Zagha, R. (2002). “From slow growth to slow reform”. Unpublished paper. World Bank. World Bank (1993). The East Asian Miracle: Economic Growth and Public Policies. World Bank, Washington, DC. World Bank (1998). Beyond the Washington Consensus: Institutions Matter. World Bank, Washington, DC. Yanikkaya, H. (2003). “Trade openness and economic growth: A cross-country empirical investigation”. Journal of Development Economics 72 (October), 57–89. Young, A. (1992). “A tale of two cities: Factor accumulation and technical change in Hong Kong and Singapore”. In: NBER Macroeconomics Annual. MIT Press for NBER, Cambridge, MA. Young, A. (2000). “The razor’s edge: Distortions and incremental reform in the People’s Republic of China”. NBER Working Paper No. 7828 (August).
Chapter 15
NATIONAL POLICIES AND ECONOMIC GROWTH: A REAPPRAISAL* WILLIAM EASTERLY New York University
Contents Abstract Keywords 1. Theoretical models that predict strong policy effects 2. Models that predict small policy effects on growth 3. Empirics 4. Some empirical caveats 5. New empirical work 6. Policy episodes and transition paths 7. Institutions versus policies 8. Conclusions References
1016 1016 1017 1026 1032 1033 1036 1050 1054 1056 1056
* I am grateful from comments received at seminars at Pompeu Fabra, Boston University, and Brown University, and in the Handbook of Economic Growth conference at New York University.
Handbook of Economic Growth, Volume 1A. Edited by Philippe Aghion and Steven N. Durlauf © 2005 Elsevier B.V. All rights reserved DOI: 10.1016/S1574-0684(05)01015-4
1016
W. Easterly
Abstract The new growth literature, using both endogenous growth and neoclassical models, has generated strong claims for the effect of national policies on economic growth. Empirical work on policies and growth has tended to confirm these claims. This paper casts doubt on this claim for strong effects of national policies, pointing out that such effects are inconsistent with several stylized facts and seem to depend on extreme observations in growth regressions. More modest effects of policy are consistent with theoretical models that feature substitutability between the formal and informal sector, have a large share for the informal sector, or stress technological change rather than factor accumulation.
Keywords economic growth, macroeconomic policies, international trade, economic reform, economic development JEL classification: O1, O4, E6, F4
Ch. 15:
National Policies and Economic Growth
1017
An influential study by World Bank researchers Paul Collier and David Dollar (2001) finds that policy reform in developing countries would accelerate their growth and cut world poverty rates in half. They conclude that Poverty reduction – in the world or in a particular region or country – depends primarily on the quality of economic policy. Where we find in the developing world good environments for households and firms to save and invest, we generally observe poverty reduction. I find the audacious claim that policy reform can cut world poverty in half a little daunting – even more so since Collier and Dollar base their results on an unpublished growth regression by me! (Like firearms, it is dangerous to leave growth regressions lying around.) The International Monetary Fund (2000) also claims that “Where {sound macroeconomic} policies have been sustained, they have raised growth and reduced poverty.” These claims are often held out as hope to economically troubled continents like Africa: “Policy action and foreign assistance . . . will surely work together to build a continent that shows real gains in both development and income in the near future.” Unfortunately, this claim was made in World Bank (1981) and the “real gains” in Africa have yet to arrive as of 2003. Do the ambitious claims for the power of policy reform find support in the data? Are they consistent with theoretical views of how policy would affect growth? The large literature on the determinants of economic growth, beginning with Romer (1986), has intensively studied national economic policies as key factors influencing long run growth. In this chapter, I take a look the state of this literature today, both theoretical and empirical. I do not claim to comprehensively survey the literature. I focus the chapter on the question of how strong is the case that national economic policies (by which I mean mainly macroeconomic and trade policies) have economically large effects on the growth rate of economies. I am in the end skeptical that national policies have the large effects that the early growth literature claimed, or that the international agencies claim today. Although extremely bad policy can probably destroy any chance of growth, it does not follow that good macroeconomic or trade policy alone can create the conditions for high steady state growth. 1. Theoretical models that predict strong policy effects The simplest theoretical model of endogenous growth is the AK mode of Rebelo (1991). Rebelo postulated that output could be proportional to a broad concept of capital (K) that included both physical and human capital: Y = AK.
(1)
In principle, K could also include any kind of stock of knowledge, technology, or organizational technique that can be built up over time by sacrificing some of today’s
1018
W. Easterly
consumption to accumulate such a stock. For example, technological knowledge could be accumulated by diverting some of today’s output into lab equipment or other machines that help make new discoveries feasible. Or knowledge or human capital itself could be used to create further knowledge or human capital rather than producing today’s output.1 However, unlike many other endogenous growth models that explicitly address knowledge or technology [e.g. Aghion and Howitt (1998)], K is treated in this model as a purely private good – both excludable and rival. I will address below what happens when we relax this assumption.2 Constant returns to the factors that can be accumulated is also a key assumption in this model’s prediction of a constant steady state rate of growth for given parameters and policies. This would rule out fixed costs in implementing a new technology, or increasing returns to accumulation at low levels of K, both of which feature in other growth models. Since K is purely a private good, there is no role for government in this model. The market equilibrium yields the first best solution, and any government intervention in the form of taxes or price distortions must worsen welfare. In this model, policies like tax rates have large effects on steady state growth. Consider first a tax (τ ) on the purchase of investment goods (I ). Consumption (C) is given by output less investment spending and taxes: C = Y − (1 + τ )I.
(2)
Suppose the population size is constant and each (identical) household-dynasty maximizes welfare over an infinite horizon: ∞ C 1−σ dt, e−ρt max (3) 1−σ 0 K˙ = I − δK. (4) 1 Rebelo (1991) showed that as long as the capital formation function itself has constant returns to accumulated factors, endogenous growth is possible even if final production has diminishing returns to capital. 2 Since K in my models can always represent either technology or factor accumulation, I do not address the hot debate on how much factor accumulation matters for growth. On education, Benhabib and Spiegel (1994) and Pritchett (1997) show that cross-country data on economic growth rates show that increases in human capital resulting from improvements in the educational attainment of the work force have not positively affected the growth rate of output per worker. It may be that, on average, education does not effectively provide useful skills to workers engaged in activities that generate social returns. There is disagreement, however, Krueger and Lindahl (2001) argue that measurement error accounts for the lack of a relationship between growth per capita and human capital accumulation. Hanushek and Kimko (2000) find that the quality of education is very strongly linked with economic growth. However, Klenow (1998) demonstrates that models that highlight the role of ideas and productivity growth do a much better job of matching the data than models that focus on the accumulation of human capital. More work is clearly needed on the relationship between education and economic development. On physical capital accumulation, there is the debate between the “neoclassical” school stressing factor accumulation [Mankiw, Romer and Weil (1992), Mankiw (1995), Young (1995)] and the school stressing technology or the residual [Klenow and Rodriguez-Clare (1997a, 1997b), Hall and Jones (1999), Easterly and Levine (2001)].
Ch. 15:
National Policies and Economic Growth
1019
Then the consumer–producer would invest at a rate that results in steady-state growth of C˙ (A/(1 + τ )) − δ − ρ (5) = . C σ Here policy has large effects on steady state growth. If A = 0.15 and σ = 1, then an increase from a tax rate of 0 to one of 30% would lower growth by 3.5 percentage points. Such a policy pursued over 30 years would leave income at the end 65 percent lower than it would have been in the absence of a tax. This is a strong claim for the effects of policy on economic development! It offers a possible explanation for the poverty of a poor nation – bad government policies (high τ ) – which can be remedied easily enough by changing to good policies (low τ ). It is clear why this has been a seductive theory for aid agencies and policymakers that seek to promote economic development. The effects on accumulation are even more dramatic. Solving for the broad concept of investment that includes physical capital, human capital, technology, and knowledge accumulation, we get: (A/(1 + τ )) − δ(1 − σ ) − ρ I = . (6) Y σA The effect of taxation on investment does not depend on A. If σ = 1, the derivative of I /Y with respect to the tax factor 1/(1 + τ ) is unity. An increase of the tax rate from 0 to 30 percent would reduce investment by 23 percentage points of GDP! Before examining this claim in more detail, note that the tax rate on investment goods does not have to be an explicit tax on capital goods. First of all, there is an equivalent income tax that would have had the same effect on growth (given by t = 1 − 1/(1 + τ )), so policies here could be any government action that diverts income away from the original investor in production. (Note using the result above, that every one percentage point increase in the income tax rate reduces investment by one percentage point of GDP.) Second, note that this result applies to the marginal effective tax rate on investment goods or income. While movements from 0 to 30 percent would be dramatic for average tax rates, a movement of 30 percentage points in marginal effective tax rates could easily come from a tax reform. Second, the tax on capital goods could stand for any policy that alters the price of investment goods relative to consumption.3 For example, suppose that a populist government controls output prices for consumers but the investor must buy goods for investment on the black market. Then the premium of the black market price over the official price would act much like a tax on investment goods. If the one good in this model is tradeable, then the black market premium on foreign
3 Chari, Kehoe and McGrattan (1996) and McGrattan and Schmitz (1998) present models and empirical work emphasizing the measured high relative price of capital goods as a policy factor inhibiting economic development. Hsieh and Klenow (2005) have an alternative story that stresses high capital prices and low income as the joint outcome of a technological disadvantage in producing tradeable goods (including capital goods) in poor countries.
1020
W. Easterly
exchange might be a good proxy for the wedge between official output prices and black market investment good prices (assuming that consumer goods can be imported at the official exchange rate, or at least that official output prices are controlled as if they could be). If we suppose that the purchaser of investment goods must hold cash in advance of a purchase of investment goods, then inflation would be indirectly be a tax on investment goods. One could also get similar results with institutional variables – a probability of expropriation of part or all of the capital good by the government or government officials demanding a bribe every time a new unit of capital is installed would act much like a tax on investment. The claims for large policy effects become even stronger in growth models with increasing returns to capital and externalities. Suppose that there is a group of large but fixed size where the capital held by each member of the group has non-pecuniary externalities for the rest of the group. For example, a high human capital individual in a residential neighborhood might benefit the rest of the neighborhood with whom she socially interacts. The knowledge and connections that this individual brings might raise the productive potential of others (this is loosely what is called “social capital” in the literature). If this is true for all social interactions in the neighborhood, and these interactions are identical, costless, and exogenous for all members, then there will be a spillover from the average human capital of the neighborhood to each inhabitant of the neighborhood. The production function for an individual member would look like this: y = Ak α k¯ β .
(7)
One can think of other similar examples of spillovers. If k includes knowledge or technology, it is plausible that these goods are non-rival and partially non-excludable. For example, firms may benefit by example from new technology installed by other firms in the same trade. People in almost every human activity engage in “shop talk” that is incomprehensible to outsiders, but which apparently conveys productive knowledge to those involved in the activity.4 Assuming the same maximization problem as above (Equations (2)–(4)), then the individual will invest in k taking everyone else’s investment as given (because the group is too large for her to influence its average). The optimal path of consumption is now given by C˙ (Aαk α−1 k¯ β /(1 + τ )) − δ − ρ (8) = . C σ However, since all members of the group are assumed to be identical, then k = k¯ expost, and the growth rate for each individual will be C˙ (Aα k¯ α+β−1 /(1 + τ )) − δ − ρ = . C σ
(9)
4 The emphasis on the special properties of knowledge and technology was highlighted by Romer (1995) and Aghion and Howitt (1998). The idea of social capital has been stressed by authors such as Putnam (1993, 2000), Glaeser, Laibson and Sacerdote (2002), Narayan and Pritchett (1997), Woolcock and Narayan (2000).
Ch. 15:
National Policies and Economic Growth
1021
There are multiple equilibria if α + β − 1 > 0, i.e. if both the original importance of broad capital to production is large plus there are strong spillovers. If we have the special case of α+β = 1, then we are back to the AK model, albeit one with suboptimal market outcomes because of the externality. If α+β −1 < 0, then the model will feature similar prediction as the neoclassical model with a high capital share (discussed below). In the multiple equilibria case, the return to capital increases the more initial capital there is, the opposite of the usual diminishing returns to capital. Figure 1 illustrates the possible outcomes. If the tax rate is low, the after tax rate of return to capital is the upper upward-sloping line. Any initial capital stock to the left of point A (where the after tax return is less than δ + ρ) will go into a vicious circle of negative growth of consumption and decumulation of capital. Any point to the right of A (such as B) will go into a virtuous circle of positive and accelerating growth of consumption and positive capital accumulation.5 Now suppose that tax rates are increased, shifting the rate of return to the lower upward-sloping line in Figure 1. Now any point to the left of C will go into a vicious circle of decline. An economy with capital stock B, which was in the expanding region under low taxes, is now in the declining region under high taxes. A policy shift now has an even more dramatic impact on national prosperity – it could spell the difference between subsistence consumption (say Mali) and industrialization (say Singapore). Policy spells the difference in the long run between per capita income of $300 and $30,000 – rather a dramatic effect. As in all multiple equilibria models, initial conditions matter and small things (like policy) can have large consequences. If the first endogenous growth model was seductive to policymakers, this is even more so – one government official at the stroke of a pen could change a nation’s prospects from destitution to prosperity. This increasing returns model is much like poverty trap models like those of Azariadis and Drazen (1990), Becker, Murphy and Tamura (1990), Kremer (1993), and Murphy, Shleifer and Vishny (1989). It is also consistent with models of in-group ethnic and neighborhood externalities [Borjas (1992, 1995, 1999), Bénabou (1993, 1996)] and geographic externalities [Krugman (1991, 1995, 1998), Fujita, Krugman and Venables (1999)]. Ades and Glaeser (1999) present evidence for increasing returns in closed economies. A story like that told in Figure 1 would also predict instability of growth rates if an economy is in the middle region B and is subject to continuous fluctuations in policies. The economy would keep shifting from positive to negative growth and back again as policies change. This is a possible story for some of the spectacular reversals in output growth that we have seen in countries like Cote d’Ivoire, Jamaica, Guyana, and Nigeria (see Figure 2). It is often assumed that these strong claims for policy effects on growth are only a feature of endogenous growth models. However, the other innovation in the growth 5 The feature of ever accelerating growth in this model leads to nonsensical predictions in the long run – the model would have to be modified at higher incomes with some feature that puts a ceiling on the rate of return to capital.
1022
W. Easterly
Figure 1. Multiple equilibria with increasing returns to capital, alternative tax regimes.
literature of the last two decades has been to put a much higher weight on capital even in the neoclassical exogenous growth model. Again, the justification is that capital is a broader concept than just physical equipment and buildings. It should include at least human capital, if not the more technology and knowledge forms of capital discussed above. Attributing part of the labor income in the national accounts to human capital, this would raise the share of capital in output from around 1/3 (if the only form of capital was physical) to something like 2/3.6 The high capital share is also necessary to avoid counterfactual predictions about very high returns to capital in capital-scarce countries, and the same in the initial years of a transition from capital-scarcity to capitalabundance. The neoclassical production function with labor-augmenting technological change is: Y = K α (AL)1−α .
(10)
In per capita terms, we have: y = k α A1−α .
(11)
The consumer–producer’s maximization problem is the same as before, using Equations (2)–(4). Technological progress (the percent growth in A) is assumed to take place
6 Mankiw, Romer and Weil (1992) and Mankiw (1995).
Ch. 15:
National Policies and Economic Growth
1023
Figure 2. Examples of variable per capita income over time.
at an exogenous rate x. As is well known, accumulation of physical and human capital cannot sustain growth in the long run in the absence of technological progress. Since policy affects the outcome only through the incentive to accumulate capital, it follows that policy by itself cannot foster sustained growth in this model. With growth in A
1024
W. Easterly
of x, the long-run steady state will have per capita output y, capital per worker k, and per capita consumption all growing at the same (exogenous) rate x. The tax rate on capital goods has no effect on the steady-state growth rate. However, policy does have potentially large effects on the level of per capita income. To see this, it is convenient to write both capital per worker and per capita income relative to the technological level A. The optimal growth of per capita consumption is now: C˙ (α(k/A)α−1 /(1 + τ )) − δ − ρ = . C σ
(12)
Since (12) must equal x in steady state, an increase in the tax rate τ must always be offset by a decrease in the relative capital stock (raising the pre-tax rate of return to capital because of diminishing returns, i.e. because α < 1). Setting (12) equal to x determines the k/A ratio in the steady state, which in turn gives the following for per capita income relative to technology: α 1−α y α = . A (1 + τ )(σ x + δ + ρ)
(13)
A high tax on investment inhibits capital accumulation and thus lowers the level of income relative to the technology level. High taxes are still a possible explanation of relative poverty in the neoclassical model. With a capital share of 2/3 (including both human and physical capital), a tax rate decrease from 50 percent to zero raises income by a factor of (1.5)2 , or 2.25 times. If the capital share were 0.8 (as writers like Barro and Mankiw have suggested), then the tax reduction would raise income by a factor of (1.5)4 or 5 times. Although there is no effect of the tax change on steady state growth, there will be a dramatic change in growth in the transition from one policy regime to another. There is one unique saddle path to the new steady state; consumption will jump to that saddle path after the change in policy (in a world of perfect certainty of course). To solve for the transition involves solving for the saddle-path of consumption in transition to the new steady state. Figure 3 shows a simulation of a decrease in the tax rate on investment from 50 percent to zero, with the following parameter values: α = 0.6666, δ = 0.07, ρ = 0.05, σ = 0.9, x = 0.02. For comparison, I also show a simulation of an endogenous growth rate model with A = 0.138, which gives the same 2 percent per capita growth rate at zero tax as the exogenous growth neoclassical model. Both models show dramatic growth rate effects after the policy change, still large after 20 to 30 years. It is only in the very long run that the neoclassical growth effect wears off with diminishing returns. Investment rates would show similar jumps after the policy change as growth rates.
Ch. 15:
National Policies and Economic Growth
1025
Figure 3. Endogenous growth and neoclassical growth with a reduction of tax rate on investment from 50 percent to zero.
1026
W. Easterly
What is different for the purposes of empirical work is that the predicted difference in growth rates in the endogenous growth model before and after the tax decrease could equally apply to cross-section differences in growth between high-tax and low-tax countries. In the neoclassical model, the predicted effect of policy change on growth is only for a cross-time effect within countries. However, this difference has been handled in practice by testing the effect of current policies on growth, controlling for initial income. Initial income can be thought of as representing policy regimes prior to the period under study. If current policy predicts a higher steady state level of income than initial income, then the transitional dynamics like those shown in Figure 3 will be set in motion. The neoclassical model would predict instability of growth rates over time if frequent policy changes shift the steady state level above or below the current income level, which is ironically similar to the increasing returns prediction of growth rate instability. One big difference between the three models is that the neoclassical model predicts falling growth and investment after the initial policy-induced increase in growth, the increasing returns to capital model predict rising growth and investment afterwards, while the constant returns to capital model predict constant growth. I will examine some case studies of major policy reforms below to see which of these predictions appears to hold. All of the three models predict large growth effects of policy changes. I will examine below the evidence for or against these claims, but here I will note how much these bold predictions are different from many other fields of economics, as well as from the pre-1986 growth literature. The literature on tax policy, for example, thinks that it is a big deal to identify a benefit of 0.1 percent of GDP from a major tax reform that lowers distortions. The notion that economic development of a whole society can be achieved a few stroke-of-the pen policy reforms seems simplistic in retrospect. If this is so, why haven’t more countries successfully developed? Are large policy effects on growth an inevitable feature of new growth models?
2. Models that predict small policy effects on growth To begin to understand some of the factors that might mitigate the large effects of policy on growth, suppose that there output is a function of two types of capital, only one of which can be taxed. For example, suppose that the first type of capital (K1 ) is formal sector capital that must be transacted on markets in the open, while the second type of capital (K2 ) is informal sector capital that can be accumulated away from the prying eyes of the tax man. γ γ1 Y = A αK1 + (1 − α)K2 γ , C = Y − (1 + τ )I1 − I2 , K˙ 1 = I1 − δK1 ,
(14) (15) (16)
Ch. 15:
National Policies and Economic Growth
K˙ 2 = I2 − δK2 ,
1027
(17) 1−γ γ
(Aα[α + (1 − α)(K2 /K1 )γ ] /(1 + τ )) − δ − ρ C˙ = . (18) C σ If these two capital goods are close to perfect substitutes, then the effects of taxes on growth go towards zero. Figure 4 shows the relationship between growth and tax rates at extreme values of γ . With γ close to 1 (close to perfect substitutability), there is only a minor effect of taxes and it is bounded from below no matter how high the tax rate. This is because with the elasticity of substitution greater than one, formal sector capital is not essential to production. The worst that high tax rates can do is drive formal capital use down to zero (which has only a small effect if the capital goods are close to perfect substitutes). After that, increases in tax rates have no further effect (explaining the flat segment of the curve in Figure 4). The effects of tax rates on growth continue to be strong if the elasticity of substitution between the two goods is less than one (the γ = −1 line in Figure 4). The other parameter that plays an important role in how damaging are tax rates is the share (α) of formal sector capital (or more specifically, the share of the capital that is actually subject to taxation). Figure 5 shows how different are the effects of taxing investment in this factor when its share (α) is 0.1 compared to when its share is 0.8 (assuming an elasticity of substitution of unity). Of course, lowering the share of taxable capital would also limit the power of taxation in the neoclassical model. Another factor that mitigates the effects of policies on growth is that many policies distort relative prices amongst different sectors or different types of goods, rather than penalizing all capital goods. With a distortion of relative prices, some capital goods are more expensive but others are cheaper. For example, with a black market premium on foreign exchange, those who receive licenses to import goods at the official exchange rate receive a subsidy, while those who must pay the black market rate for inputs pay an implicit tax.7 Unanticipated high inflation is a tax on creditors but a subsidy to debtors. An overvalued real exchange rate penalizes producers of tradeables but subsidizes producers of nontradeables. Trade protection taxes imports but subsidizes production for the domestic market. The rate of subsidy is clearly related to the rate of taxation. One way to pin it down is to specify that the revenues from the tax on the first type of capital must just cover the subsidy expenditures on the second type of capital. Here are the equations I have in mind. I revert to Cobb–Douglas for simplicity: Y = AK1α K21−α ,
(19)
C = Y − (1 + τ )I1 − (1 − s)I2 .
(20)
(16) and (17) still represent the capital accumulation equations, and the consumerproducer maximizes (3) taking τ and s as given. Ex-post, the government must balance 7 If black markets function efficiently, the opportunity cost of inputs is their black market value even for those who receive them at the subsidized price. However, the recipient of inputs at the official exchange rate still receives a subsidy per unit of input use.
1028
W. Easterly
Figure 4. Growth rates with different assumptions about elasticity of substitution between capital good types.
Ch. 15:
National Policies and Economic Growth
1029
Figure 5. Tax rates and growth with different shares of taxable capital.
its budget so: τ I1 = sI2 .
(21)
Because of the neat properties of Cobb–Douglas, the solution of the optimal capital ratio as a function of the subsidy rate (after taking into account the fiscal relation-
1030
W. Easterly
ship (21) between tax rates and subsidy rates) is very simple: K2 1−α . = K1 α−s
(22)
The growth rate will display offsetting effects of the subsidy-cum-tax rate – on the one hand, it distorts the allocation of capital away from K1 to K2 , lowering the presubsidy marginal product of K2 , while on the other hand, it of course subsidizes the rate of return to K2 . C˙ (A(1 − α)((α − s)/(1 − α))α /(1 − s)) − δ − ρ (23) = . C σ One can show that if (21) (the balanced budget requirement) is imposed, it is impossible for this kind of tax-cum-subsidy scheme to raise the rate of growth.8 The tax-cum-subsidy will imply an efficiency loss from the distortion of resource allocation, and this efficiency loss will have a negative growth effect if all types of capital can be accumulated. However, the relationship between the distortion and the growth rate is highly nonlinear. As is well known in the literature on relative price distortions, the cost of the distortion increases more than proportionately with the size of the distortion.9 In the traditional literature on “Harberger triangles”, this was an output loss. In an endogenous growth model where all inputs can be accumulated, the distortion between relative prices of the inputs induces a reduction in growth. A small distortion introduces only a small wedge in between marginal products of the two inputs and does not cause a huge growth loss. Eventually, however, the distortion forces far too much accumulation of one type of capital relative to the other, severely lowering the marginal product of the excessive capital good due to diminishing returns. An increasing rate of subsidy also requires a more than one for one increase in the tax rate, as the tax base is shrinking with increased taxes while the capital goods being subsidized are increasing. The nonlinear relationship is shown in Figure 6. Note that distortions do not have much effect on growth at all up to subsidy rates of about 0.2 and then have increasingly catastrophic consequences after about 0.4. There are other factors that mitigate the effects of policy on growth that I do not explicitly model here. One is policy uncertainty. The announcement of a new policy may not be credible, perhaps because high political opposition to it may imply a high probability of subsequent reversal. Many developing countries have a history of frequent reversals of incipient policy reforms, which makes any future reform less believable. For example, Argentina has been a chronic high inflation country for nearly half a century. Frequent stabilization attempts have subsequently come unwound; the fiasco of the Convertibility Plan in 2001 is only the latest example. In terms of the model above, the
8 This applies to CES production functions more generally [see Easterly (1993) for a proof]. 9 One recent growth model emphasizing this nonlinearity is Gylfason (1998), where the cost e of a distortion c is amusingly expressed as e = mc2 .
Ch. 15:
National Policies and Economic Growth
1031
Figure 6. Growth rate and subsidy rate financed by taxes.
certainty equivalent of the after-tax return on capital may not increase much even after an announcement that taxes will be cut. There is also the possibility that policies whose main purpose was to create rents for political patronage will be replaced with other policies that create new rents. For example, if the black market premium is abolished, the holders of import licenses at the
1032
W. Easterly
official exchange rate may seek new sources of income (for example, appointment as customs inspectors, where they can take bribes). There may be a law of conservation of political rents, akin to the second law of thermodynamics, if the factors inducing political rent seeking do not change. Poor countries may be so close to subsistence consumption that they may not be able to take advantage of policy changes. Rebelo (1992) and Easterly (1994) show intertemporal utility functions with Stone–Geary preferences, in which consumers derive utility from consumption only above a certain floor of subsistence. This model predicts a very low intertemporal elasticity of substitution at levels of consumption close to subsistence. Intuitively, consumers close to subsistence have a limited ability to postpone consumption in order to take advantage of higher returns to saving. This model predicts a slow acceleration of growth even after a favorable policy change, as consumption must first rise well above subsistence. Most importantly, policies may be offset or reinforced by more important factors that affect the growth and income. Achieving high output returns from a given set of inputs involves an incredibly complex set of institutions (such as enforcement of contracts and property rights), social norms, efficient sorting and matching of people and other inputs, advanced technological knowledge, full information on both sides of all transactions, low transaction costs, resolution of principal-agent problems, positive non-zero-sum game theoretic interactions among agents, resolution of public good problems, and so on. The development of institutions and social and political structures that address these issues successfully (from the standpoint of material production) is probably a long historical process. The above models have a pale shadow of all this complexity in the parameter A. Note that the lower is A, the lower is the derivative of growth (or income in the neoclassical model) with respect to the policy parameter τ . Many authors have argued that differences in A explain a large part of income differences between countries [Hall and Jones (1999), Klenow and Rodriguez-Clare (1997a, 1997b), Easterly and Levine (2001)]. If a poor country is poor because of low A, then a change in policies may not do much to raise income or growth. Exogenous variation in A may also affect the political economy of policy – a high A country would be less likely to tolerate the costs of destructive policies, while bad policy may be tolerated in a low A country because it may not make much difference. Of course policy itself could influence A. However, if A really depends on all the complexities listed above, then the kind of macroeconomic policies I am considering in this paper may not have much effect on A.
3. Empirics The literature tracing effects of economic policies on growth is abundant. I do not attempt to summarize it here, noting the summaries in Sala-i-Martin (2002), Temple (1999), Kenny and Williams (2001), and Easterly and Levine (2001). Some authors focus on openness to international trade [Frankel and Romer (1999)], others on fiscal
Ch. 15:
National Policies and Economic Growth
1033
policy [Easterly and Rebelo (1993a, 1993b)], others on financial development [Levine, Loayza and Beck (2000)], and others on macroeconomic policies [Fischer (1993)]. Dollar (1992) stressed a measure of real exchange rate overvaluation as a proxy for outward orientation and thus a determinant of growth. These papers have at least one common feature: they all find that some indicator of national policy is strongly linked with economic growth, which confirms the argument made by Levine and Renelt (1992) – even though Levine and Renelt found that it was difficult to discern which policy matters for growth. The list of national economic policies that have received most extensive attention are fiscal policy, inflation, black market premiums on foreign exchange, financial repression vs. financial development, real overvaluation of the exchange rate, and openness to trade. The recommendation that countries pursue good policies on all these dimensions was labeled by Williamson (1990) as the “Washington Consensus”. I distinguish policies from “institutions”, which have their own rich literature [see Acemoglu, Johnson and Robinson (2001, 2002), La Porta et al. (1999, 1998), Kaufmann, Kraay and Zoido-Lobatón (1999), Levine (2005)]. Institutions reflect deepseated social arrangements like property rights, rule of law, legal traditions, trust between individuals, democratic accountability of governments, and human rights. Although governments can slowly reform institutions, they are not “stroke of the pen” reforms like changes in the macroeconomic policies listed above. I will consider at the end the relative role of policies and institutions in development.
4. Some empirical caveats There are several things to note about the evidence on policies and growth before proceeding to new empirical analysis. The first is that the literature has devoted much effort to the most obvious candidate for a policy that influences growth – tax rates. Yet the literature has generally failed to find a link between income or output taxes and economic growth [Easterly and Rebelo (1993a, 1993b), Slemrod (1995)]. Nor are we likely to find that taxes have level effects, as rich countries have higher tax rates than poor countries. The outcome of natural experiments like the large tax increases in the US associated with the introduction of the income tax and the World Wars does not indicate income or level effects of taxes [Rebelo and Stokey (1995)]. Hence, the most obvious policy variable affecting growth is out of the running from the start. Second, national economic policies are generally measured over the period 1960– 2000, which is when data is available. This is also the period in which countries had independent governments making policy, as opposed to colonial regimes (on which we do not have data). Hence, if policies have an effect on the level or growth rate of income, this would have to show up in the period 1960–2000. However, history did not begin with a clean slate in 1960. The correlation of per capita income in 1960 with per capita income in 1999 is 0.87. Most of countries’ relative performance is explained by the point they had already reached by 1960. It follows that the role of post-1960 policies
1034
W. Easterly
Figure 7. Predicted vs actual per capita growth for developing countries (assuming constant intercept across decades).
in determining development outcomes can only be limited. A view of economic development that puts all the weight on the 1960–2000 period is ahistorical, assuming away the complex histories of civilizations, conquests, and colonies. Third, there is the general fact that developing countries had higher growth rates in the period 1960–1979 than in the period 1980–2000. Yet most of the “Washington Consensus” policies were adopted only after 1980. In the pre-1980 days, there was much more of an emphasis on state intervention and import-substituting industrialization, as opposed to the free trade, “get the prices right” approach after 1980. This big fact does not augur well for a strong positive effect of “good policies” on growth, although the growth slowdown after 1980 could have other causes. Easterly (2001) showed the divergence between improving growth predicted by policies and actual growth outcomes across the 60s, 70s, 80s, and 90s (see Figure 7). Fourth, there are many income differences within nations – between the sexes, between ethnic groups, and between regions – that cannot be explained by national economic policies. Easterly and Levine show that there are four ethnic–geographic clusters of counties with poverty rates above 35 percent in the US: (1) Counties in the West that have large proportions (>35%) of native Americans; (2) Counties along the Mexican border that have large proportions (>35%) of Hispanics; (3) Counties adjacent to the lower Mississippi river in Arkansas, Mississippi, and Louisiana and in the “black belt” of Alabama, all of which have large proportions of blacks (>35%); (4) Virtually all-white counties in the mountains of eastern Kentucky. The county data did not pick up the well-known inner-city form of poverty, mainly among blacks, be-
Ch. 15:
National Policies and Economic Growth
1035
cause counties that include inner cities also include rich suburbs. An inner city zip code in DC, College Heights in Anacostia, has only one-fifth of the income of a rich zip code (20816) in Bethesda MD. This has an ethnic dimension again since College Heights is 96 percent black and the rich zip code in Bethesda is 96 percent white. The purely ethnic differentials in the US are well known. Blacks earn 41 percent less than whites; Native Americans earn 36 percent less; Hispanics earn 31 percent less; Asians earn 16 percent more.10 There are also more subtle ethnic earnings differentials. Thirdgeneration immigrants with Austrian grandparents had 20 percent higher wages in 1980 than third-generation immigrants with Belgian grandparents [Borjas (1992)]. Among Native Americans, the Iroquois earn almost twice the median household income of the Sioux. Other ethnic differentials appear by religion. Episcopalians earn 31% more income than Methodists [Kosmin and Lachman (1993), p. 260]. Twenty-three percent of the Forbes 400 richest Americans are Jewish, although only two percent of the US population is Jewish [Lipset (1996)].11 Poverty areas exist in many countries: northeast Brazil, southern Italy, Chiapas in Mexico, Balochistan in Pakistan, and the Atlantic provinces in Canada. Bouillon, Legovini and Lustig (2003) find that there is a negative Chiapas effect in Mexican household income data, and that this effect has gotten worse over time. Households in the poor region of Tangail/Jamalpur in Bangladesh earned less than identical households in the better off region of Dhaka [Ravallion and Wodon (1998)]. Ravallion and Jalan (1996) and Jalan and Ravallion (1997) likewise found that households in poor counties in southwest China earned less than households with identical human capital and other characteristics in rich Guangdong Province. In Latin America, the main ethnic divide is between indigenous and non-indigenous populations and between white, mestizo, and black populations. In Mexico, 80.6 percent of the indigenous population is below the poverty line, while only 18 percent of the nonindigenous population is below the poverty line.12 But even within indigenous groups in Latin America, there are ethnic differentials. There are 4 main language groups among Guatemala’s indigenous population. Patrinos (1997) shows that the Quiche-speaking indigenous groups in Guatemala earn 22 percent less on average than Kekchi-speaking groups. In Africa, there are widespread anecdotes about income differentials between ethnic groups, but little hard data. The one exception is South Africa. South African whites
10 Tables 52 and 724, 1995 Statistical Abstract of US. 11 Ethnic differentials are also common in other countries. The ethnic dimension of rich trading elites is well-
known: the Lebanese in West Africa, the Indians in East Africa, and the overseas Chinese in Southeast Asia. Virtually every country has its own ethnographic group noted for their success. For example, in the Gambia a tiny indigenous ethnic group called the Serahule is reported to dominate business out of all proportion to their numbers – they are often called “Gambian Jews”. In Zaire, Kasaians have been dominant in managerial and technical jobs since the days of colonial rule – they are often called “the Jews of Zaire” (New York Times, 9/18/1996). 12 Source: Psacharopoulos and Patrinos (1994, p. 6).
1036
W. Easterly
Figure 8. Persistence over time of policies and growth.
have 9.5 times the income of blacks. More surprisingly, among all-black traditional authorities (an administrative unit something like a village) in the state of KwaZuluNatal, the ratio of the richest traditional authority to the poorest is 54 [Klitgaard and Fitschen (1997)]. While not ruling out national policy effects, these differences also highlight the importance of factors that do not operate at the national level. Fifth, the role of policies in explaining post-1960 growth is bounded once we realize that policy variables are much more stable over time than are growth rates.13 Figure 8 shows the correlation coefficient across successive 5-year periods between different kinds of policies and growth. As noted in the theoretical section, stability of policies over time and instability of growth rates is inconsistent with the AK model. It could be consistent with either the neoclassical model or the increasing returns growth model, assuming that policies are close to the steady state or critical point, respectively. Note that the non-persistence of growth rates and the high persistence of income levels is consistent, since persistent differences in growth rates would be required to scramble the income rankings from 1960 to 1999. 5. New empirical work I here synthesize past results by running new regressions on an updated dataset for the years 1960–2000, using a panel of five year averages. Following the literature, I con13 This was pointed out by Easterly et al. (1993).
Ch. 15:
National Policies and Economic Growth
1037
Table 1 Variables used in analysis Variable name
Definition
Source
LGDPG
Log per capita growth rate
World Bank (2002)
INFL
Log (1 + inflation rate)
World Bank (2002)
BB
Government budget balance/GDP
World Bank (2002)
M2
M2/GDP
World Bank (2002)
LREALOVR
Log (overvaluation index/100) (above zero indicates overvaluation)
World Bank (2002)
LBMP
Log (1 + black market premium on foreign exchange)
World Bank (2002)
TRADE
(Exports+Imports)/GDP
World Bank (2002)
GOVC
Government consumption/GDP
World Bank (2002)
PRIV
Private sector credit/total credit
World Bank (2002)
LNEWGDP
Log of per capita GDP
Summers and Heston (1991) updated using LGDPG
LTYR
Log of total schooling years
Barro and Lee (2000)
centrate on the most common measures of macroeconomic policies, price distortions, financial development, and trade openness. My variables are listed in Table 1. Table 2 shows the variables’ summary statistics. Table 3 shows the correlation coefficients between these variables and growth as well as between distinct policies. All of the bivariate correlations of policy variables with per capita growth are statistically significant at the 5 percent level. Most of the pairwise correlations between policy variables are also statistically significant, indicating the problem of collinearity that has plagued the literature. Bad policies tend to go together along a number of dimensions. M2 and PRIV have such a high correlation that it is clear they are measuring the same thing – the overall level of financial development.
1038
W. Easterly Table 2 Summary statistics
Variable
Number of observations
Mean
Standard deviation
Min
Max
967 921 1306 1241 958 1064 916 609 1024 1270 832
0.159 8.107 0.017 15.790 −0.037 0.349 0.355 0.060 0.254 0.702 1.277
0.325 1.040 0.051 6.700 0.054 0.253 0.329 0.387 0.558 0.454 0.820
−0.569 5.775 −0.736 3.915 −0.417 0.009 0.000 −1.206 −1.058 0.018 −2.453
3.447 10.445 0.276 58.310 0.391 1.929 2.085 1.612 8.311 3.803 2.476
INFL LNEWGDP LGDPG GOVC BB M2 PRIV LREALOVR LBMP Trade LTYR
Table 3 Correlation coefficients
LGDPG INFL BB LREALOVR LBMP M2 Trade PRIV GOVC
LGDPG
INFL
BB
LREALOVR
LBMP
1.000 −0.376 0.155 −0.213 −0.321 0.097 0.101 0.130 −0.130
−0.376 1.000 −0.201 0.078 0.287 −0.193 −0.078 −0.212 0.031
0.155 −0.201 1.000 −0.141 −0.144 −0.010 0.094 0.110 −0.231
−0.213 0.078 −0.141 1.000 0.247 −0.083 −0.056 −0.028 0.228
−0.321 0.287 −0.144 0.247 1.000 −0.073 −0.178 −0.241 −0.036
M2
Trade
PRIV
GOVC
0.097 0.101 0.130 −0.130 −0.193 −0.078 −0.212 0.031 −0.010 0.094 0.110 −0.231 −0.083 −0.056 −0.028 0.228 −0.073 −0.178 −0.241 −0.036 1.000 0.375 0.716 0.246 0.375 1.000 0.161 0.276 0.716 0.161 1.000 0.215 0.246 0.276 0.215 1.000
I now concentrate on a core set of six variables that seem to capture distinct dimensions of policy: inflation, budget balance, real overvaluation, black market premium, financial depth, and trade openness. Initially, I will test the AK model’s prediction that these policies will have growth rather than level effects, so I do not control for initial income (I will check this later on). I will use a variety of specifications and econometric methods to assess how robust are the statistical associations between policies and growth. I start off with a figure emphasizing the bivariate association between growth and different policies (Figure 9). I divide the sample into two parts, picking out the minority part of the sample where policy is extremely bad and comparing it to the rest (for inflation, black market premium, real overvaluation, and budget balance). Inflation, black market premium, and budget balance all have a distribution featuring a long tail of extreme “bad policy”, which seems like a real world experiment worth investigating. So I eyeball the distribution and pick a threshold that picks out this tail of bad policy.
Ch. 15:
National Policies and Economic Growth
1039
Figure 9. Bivariate effects of policy on growth.
Trade/GDP and M2/GDP have a long tail for extremely good policy, so I pick a threshold picking out the extremes of good policy (see Figures 10–15). Real overvaluation does not have a long tail in one direction or the other, but I follow the same practice as with inflation, black market premium, and budget balance in setting a threshold that picks out extremely bad policy. Figure 9 shows that these experiments of either extremely good or extremely bad policy are associated with important growth differences. All of the differences are statistically significant except for the results on M2/GDP. Such strong associations have contributed to the conventional wisdom that policy has strong growth effects. In Table 4, I regress growth on all six policy variables, and then try dropping one at a time. In the base specification, four of the six policies are statistically significant at the 5 percent level, with trade openness just barely falling short. When I experiment with dropping one variable at a time, all of the six policy variables are significant at one time or another. The coefficients on the policy variables are fairly stable across different permutations of the variables.14 Table 5 shows the effect on growth of a one standard deviation improvement in each of the policy variables on growth. If all six variables were improved at the same time, 14 The other policy variables that I tested: government consumption and private sector credit, were not signif-
icant when entered in addition to these variables (or substituting government consumption for budget deficits and private sector credit for M2).
1040
W. Easterly
Figure 10. Histogram of inflation (truncated between 0 and 1).
Figure 11. Histogram of real overvaluation (truncated between −1 and 1).
Ch. 15:
National Policies and Economic Growth
Figure 12. Histogram of budget balance/GDP.
Figure 13. Histogram of trade/GDP (percent).
1041
1042
W. Easterly
Figure 14. Histogram of M2/GDP (percent).
Figure 15. Histogram of Log of (1 + black market premium).
Ch. 15:
National Policies and Economic Growth
1043
Table 4 Regressions of per capita growth on basic set of 6 policy variables. Dependent variable: LGDPG (log per capita growth, five year averages, 1960–2000) INFL BB M2 LREALOVR LBMP Trade Constant Observations R-squared
−0.018 (2.61)∗∗ 0.092 (2.81)∗∗ 0.01 1.37 −0.014 (2.97)∗∗ −0.012 (2.33)∗ 0.01 1.92 0.016 (3.62)∗∗ 422 0.18
−0.02 (3.13)∗∗ 0.114 (3.48)∗∗ 0.013 1.92 −0.013 (2.98)∗∗ −0.017 (3.43)∗ 0.011 (2.22)∗ 0.013 (3.09)∗∗ 434 0.15
0.014 (2.04)∗ −0.016 (3.74)∗∗ −0.01 (2.06)∗ 0.011 (2.15)∗ 0.01 (2.33)∗ 458 0.16
−0.02 (2.87)∗∗ 0.092 (3.07)∗∗
−0.013 (2.83)∗∗ −0.014 (2.73)∗∗ 0.012 (2.62)∗∗ 0.021 (5.67)∗∗ 495 0.17
−0.034 (6.27)∗∗ 0.053 (3.07)∗∗ 0.017 (2.26)∗
−0.005 −0.93 0.001 0.31 0.019 (4.81)∗∗ 573 0.13
−0.021 (3.39)∗∗ 0.109 (3.37)∗∗ 0.013 (1.99)∗ −0.015 (3.56)∗∗
0.008 (2.13)∗ 0.015 (3.92)∗∗ 455 0.17
−0.018 (2.60)∗∗ 0.098 (2.92)∗∗ 0.015 (2.15)∗ −0.013 (2.88)∗∗ −0.013 (2.60)∗∗
0.021 (5.55)∗∗ 424 0.17
Robust standard errors, significant t statistics in parentheses. ∗ Significant at 5%. ∗∗ Significant at 1%.
Table 5 Effect of one standard deviation improvement in each policy variable on economic growth Variable
INFL BB M2 LREALOVR LBMP Trade Sum
Improvement of one standard deviation in policy variable −0.325 0.054 0.253 −0.387 −0.558 0.454
Coefficient in growth regression −0.018 0.092 0.010 −0.014 −0.012 0.010
Change in growth from one standard deviation change in policy (%) 0.6 0.5 0.3 0.5 0.7 0.5 3.0
the regression suggests a 3 percentage point improvement in per capita growth. These results also seem to support the assertion that policies have strong effects on per capita growth. The promise of getting 3 additional percentage points of growth due to a moderate policy reform package is very seductive. However, there is something disquieting about
1044
W. Easterly
these results upon further reflection. The one standard deviation change in the policy variables is often very large: reduction of 0.32 in log inflation, 5 percentage point improvement in the budget balance as a ratio to GDP, 25 percentage point increase in M2/GDP, reduction of −0.39 in log real overvaluation, reduction of −0.56 in log black market premium, and increase of 45 percentage points in trade/GDP ratio. Such large changes are outside the experience of most countries with moderate inflation, budget deficits, real overvaluation, black market premiums, etc. The large standard deviations are related to the long tails I mentioned above. Except for the real overvaluation index, all of the policy variables are highly skewed, with most of the sample concentrated at low values and a few very extreme observations. The outlying observations of inflation, budget deficits, and black market premium are realizations of extreme “bad policies”. The outlying observations of trade/GDP and M2/GDP are realizations of extreme “good policies”. It is econometric commonsense that extreme observations can be very influential in determining statistical significance of right-hand side variables. How do the above regressions do over more moderate ranges of policy variables? Table 6 shows the effect of restricting the sample to observations where all six policy variables lie in the range of “moderate” policies. Moderate is defined rather arbitrarily by eye-balling the histograms above to determine where are the cutoffs containing the bulk of the sample (the same cutoffs as in Figure 9 above). Nevertheless, the cutoffs would fit a common-sense description of “extremes”: inflation and black market premiums more than 0.3 in log terms (35 percent), real overvaluation more than 0.5 (68 percent), budget deficits greater than 12 percent of GDP, M2 to GDP ratios of more than 100 percent, and trade to GDP ratios of more than 120 percent. The results of excluding any observation where any of the six policy variables are “extreme” is striking: all six policy variables become insignificant, and the F-statistic for their joint effect also falls short of significance. This is not to dismiss the evidence for policy effects on growth (reducing the range of the right-hand side variables would be expected to diminish statistical significance). These extremes are far from irrelevant, as observations in which at least one of the six policies was “extreme” account for more than half the sample. However, these results highlight the dependence of the policy and growth evidence on extreme observations of the policy variables. (The significance of extreme values and the insignificance of moderate ones is also consistent with the prediction of the theoretical model on the nonlinear effects of tax-cum-subsidy policies on economic growth.) There is also the possible endogeneity of these extreme policies, which may reflect general institutional or political chaos. The results suggest that countries not undergoing extreme values of these variables do not have strong reasons to expect growth effects of moderate changes in policies.15
15 The empirical literature on inflation has found that inflation only has a negative effect above some threshold
level, although there are disagreements as to where that threshold is [Bruno and Easterly (1998), Barro (1995, 1998), Sarel (1996)].
Ch. 15:
National Policies and Economic Growth
1045
Table 6 Robustness of results to restricting sample to moderate policy range. Dependent variable is LGDPG Sample
INFL BB M2 LREALOVR Trade LBMP Constant Observations R-squared
Full
−0.018 (2.61)∗∗ 0.092 (2.81)∗∗ 0.01 1.37 −0.014 (2.97)∗∗ 0.01 1.92 −0.012 (2.33)∗ 0.016 (3.62)∗∗ 422 0.18
Moderate policies −0.064 −1.23 0.018 0.22 −0.004 0.27 0.001 0.06 0.01 1.09 −0.038 −0.95 0.027 (2.52)∗ 193 0.03
Robust t statistics in parentheses. Restrictions under moderate policies: INFL between −0.05 and 0.3, BB between −0.12 and 0.02, M2 < 1.0, LREALOVR between −0.5 and 0.5, Trade < 1.20, LBMP between −0.05 and 0.3. ∗ Significant at 5%. ∗∗ Significant at 1%.
These results are fairly intuitive if we think of destroying growth as a different process from creating growth. It is a lot easier to cut down a tree than to grow one.16 Countries that pursue destructive policies like high inflation, high black market premium, chronically high budget deficits and other signs of macroeconomic instability are plausible candidates to miss out on growth. However, it doesn’t follow that one can create growth with relative macroeconomic stability. The policies are inherently asymmetric – a leader can sow chaos by printing money and controlling the exchange until he gets a hyperinflation and an absurd black market premium. However, the best he can do in the other direction is zero inflation and zero black market premium. The results on policies and growth may simply reflect the potential for destruction from bad policies, not the potential for fostering long run development through good policy. The only exception to this story is the trade/GDP variable, whose significance depended on “extremely good” policies. Whatever the source of the result on the extreme, 16 Easterly (2001) has a chapter “How governments can destroy growth”.
1046
W. Easterly Table 7 Results on initial income and schooling Dependent variable
LGDPG
LGDPG
LGDPG
LGDPG
INFL
−0.018 (2.61)∗∗ 0.092 (2.81)∗∗ 0.010 1.37 −0.014 (2.97)∗∗ 0.01 1.92 −0.012 (2.33)∗
−0.019 (2.67)∗∗ 0.102 (2.44)∗ 0.004 0.41 −0.014 (3.07)∗∗ −0.01 −1.83 0.01 1.87 0.003 1.4
−0.02 (2.65)∗∗ 0.124 (2.65)∗∗ 0.002 0.16 −0.013 (2.40)∗ −0.01 −1.63 0.008 1.37 −0.001 −0.28 0.007 1.42
−0.019 (2.85)∗∗ 0.107 (2.57)∗ 0.006 0.67 −0.014 (2.96)∗∗ −0.011 −1.96 0.009 1.62 0.0480 1.96
BB M2 LREALOVR Trade LBMP LNEWGDP LTYR LNEWGDP2 Constant Observations R-squared
0.016 (3.62)∗∗ 422 0.18
−0.004 −0.25 411 0.18
0.019 −0.87 359 0.19
−0.0030 −1.87 −0.187 −1.86 411 0.18
Robust t statistics in parentheses. Turning point for convergence is 2981. ∗ Significant at 5%. ∗∗ Significant at 1%.
this suggests that opening up for most economies – who likely would not reach this extreme even under complete free trade – would not be associated with growth effects. The next thing to test is whether initial income belongs in the growth equation, as the neoclassical model would imply. It has also been common in the literature to add initial schooling as an indicator of whether the balance between physical and human capital is far from the optimal level. Table 7 shows the results on initial income and schooling. The results are not very supportive of a conditional convergence result. Initial income and schooling do not enter significantly, although a nonlinear formulation of hump-shaped conditional convergence (including initial income squared) comes close to significance.17 Since there is a large literature starting with Barro (1991) and Barro and Sala-i-Martin (1992) that does find conditional convergence, I do not claim this result 17 Hump-shaped convergence is consistent with a neoclassical model in which there is some subsistence floor
to consumption (the Stone–Geary utility function).
Ch. 15:
National Policies and Economic Growth
1047
Table 8 Panel methods in policies and growth regressions. Dependent variable: LGDPG Panel method
INFL BB M2 LREALOVR Trade LBMP Constant Observations Number of country units R-squared Sample Reject random effects
Random effects
Between
Fixed effects
−0.019 (3.53)∗∗ 0.082 (2.35)∗ 0.002 −0.22 −0.009 −1.8 0.012 −1.95 −0.011 (2.15)∗ 0.017 (3.22)∗∗ 422 88 0.17
−0.012 −0.97 0.216 (3.51)∗∗ 0.026 (2.19)∗ −0.027 (3.82)∗∗ 0 −0.07 −0.01 −0.97 0.019 (2.73)∗∗ 422 88 0.41
−0.02 (3.43)∗∗ 0.069 −1.64 −0.057 (3.16)∗∗ 0.01 −1.43 0.046 (3.19)∗∗ −0.012 −1.84 0.016 −1.61 422 88 0.13
Full Yes
Full
Full
Absolute value of z statistics in parentheses. ∗ Significant at 5%. ∗∗ Significant at 1%.
is decisive. It does show the fragility of the results on both policies and initial conditions (note that three of the policy variables become insignificant when initial income is included). I will come back to the issue of conditional convergence when I examine effects of policy on growth with dynamic panel methods. There is another robustness check that we should perform on the policies and growth results. Following common practice in the literature, I have been doing regressions on pooled time series cross-section observations. This implicitly assumes that the effects on growth of a policy change over time are the same as a policy difference between countries. It is straightforward to test this restriction by doing within and between regressions on the pooled sample. Table 8 shows the results. I also show the results of a random effects regression, which gives results similar to OLS on the pooled sample. The test of whether the random effects are orthogonal to the right-hand side variables is an indirect test of the equality of the coefficients from the between and within regressions. I strongly reject the hypothesis that the random effects are orthogonal. We can see from the between and within (fixed effects) regressions that the coefficients across time and across countries are indeed very different. Inflation is not significant in the be-
1048
W. Easterly
tween regression but strongly significant in the within regression.18 The budget balance is the reverse: strongly significant in the between regression but not in the fixed effects regression. The weak result that I found on M2/GDP in the pooled regression turns out to be because the between and within effects tend to cancel out: M2/GDP is strongly positively correlated with growth in the between regression and negatively correlated with growth in the within regressions. Real overvaluation and trade also show different results in the two different panel methods (real overvaluation is significant between countries and insignificant within countries, while trade is the reverse). This instability of growth effects is inconsistent with a simple AK view of growth with instantaneous transitional dynamics. It is also possible that five year averages are not long enough to wipe out cyclical fluctuations. The negative correlation between M2/GDP and growth could be seen as a cyclical pattern such as a loosening of monetary policy during recessions and tightening during booms. Likewise the correlation of trade/GDP with growth could indicate that international trade is pro-cyclical, as opposed to indicating any causal effect of openness on growth. Also note that the r-squared of the between regression is much higher than the within regression. This is not surprising given that the between regression is on averages, but it does show that the growth effects of most concern to policy makers – the change over time within a given country of growth in response to policy changes – are very imprecisely estimated by the data. Fully 87 percent of the within country variance in growth rates is not explained by these six policy variables. This result is not surprising when we recall the persistence of policies over time and the non-persistence of growth rates. Another panel method I apply to the data is the well-known dynamic panel estimator of Arellano and Bond (1991). This estimator uses first differences to remove the fixed effects. This method has several advantages: (1) it addresses reverse causality concerns by using twice-lagged values of the right-hand side variables as instruments for the first differences of RHS variables, (2) we can include initial income again, which is not possible with traditional panel methods because it would be correlated with the error term (Arellano and Bond address this by instrumenting for initial income with the twicelagged value), and (3) we can also include the lagged growth rate to allow for partial adjustment of growth to policy changes, which is more plausible than instantaneous adjustment. The results in Table 9 are notable in reinvigorating the conditional convergence hypothesis. This is consistent with previous work that shows a higher coefficient (in absolute value) on initial income with dynamic panel methods than with pooled or crosssection OLS [Caselli, Esquivel and Lefort (1996)]. The coefficient on lagged growth is not significant, failing to find support for the partial adjustment hypothesis. The results on policies are similar (not surprisingly) to the fixed effects estimator above. Inflation
18 This is consistent with the Bruno and Easterly (1998) result that high inflation crises have a strong tempo-
rary negative effect on output but no permanent effects.
Ch. 15:
National Policies and Economic Growth
1049
Table 9 Regressions using Arellano and Bond dynamic panel method Dependent variable: LGDPG LD.LGDPG D.INFL D.BB D.M2 D.LREALOVR D.LBMP D.Trade
(1)
(2)
(3)
(4)
−0.0441 (0.0749) −0.0137 (0.0068)∗∗ 0.1014 (0.0509)∗∗ −0.0701 (0.0286)∗∗ 0.0085 (0.0093) −0.0084 (0.0086) 0.0715 (0.0201)∗∗∗
−0.1131 (0.0674)∗ −0.0141 (0.0064)∗∗ 0.0958 (0.0501)∗ −0.0457 (0.0284) 0.0081 (0.0087) −0.008 (0.0081) 0.072 (0.0193)∗∗∗ −0.0487 (0.0098)∗∗∗
−0.09 (0.0771) −0.0162 (0.0065)∗∗ 0.0876 (0.0540) −0.0522 (0.0302)∗ 0.0083 (0.0092) −0.0037 (0.0086) 0.0635 (0.0204)∗∗∗ −0.0508 (0.0104)∗∗∗ 0.0091 (0.0104) 0.0019 (0.0020)
−0.0627 (0.0823) −0.017 (0.0066)∗∗∗ 0.0544 (0.0571) −0.0486 (0.0307) 0.0021 (0.0098) 0.0013 (0.0090) 0.0555 (0.0211)∗∗∗ −0.0466 (0.0105)∗∗∗ 0.0137 (0.0105) 0.001 (0.0021)
275 69 31.19113 0.0385 0 0.5212 No
275 69 23.98194 0.1968 0 0.4797 Yes
D.LNEWGDP D.LTYR Constant
−0.0012 (0.0016)
0.0014 (0.0016)
Observations Number of country Sargan Prob > CHI2 Test first order autocovariance Test second order autocovariance Time dummies
323 82 36.51018 0.0091 0 0.9091 No
316 79 35.57537 0.0119 0 0.4666 No
Standard errors in parentheses. ∗ Significant at 10%. ∗∗ Significant at 5%. ∗∗∗ Significant at 1%.
and trade are strongly significant with the right sign, while M2/GDP still has a significant but perverse sign. The results do not change much if I experiment with omitting one policy variable at a time. The estimates are consistent because I fail to reject that second order serial correlation is zero. The difference with the fixed effects result on policies is that these results have somewhat more claim to being causal. However, the Sargan test rejects the overidentifying restrictions, except in the last equation where I add time dummies. This highlights a weakness of the strong claims for causality made by the dynamic panel method – they depend on the rather dubious assumption that the lagged right-hand side variables do not themselves enter the growth equation. The same problem afflicts the cross-section or pooled regressions that use lagged values of policy
1050
W. Easterly
as instruments for current policies. Traditionalists who like intuitive arguments why instruments plausibly affect the independent but not the dependent variable are not very persuaded by lagged policies as instruments. As Mankiw (1995) noted sarcastically, if I instrument for the price of apples with the lagged price of apples in an equation for the quantity of apples, is it the supply or demand equation that I have identified?
6. Policy episodes and transition paths A more informal approach to detecting the nature of policy effects on growth is to do episodic analysis – try to identify major policy reforms and simply examine what happened to growth and investment before and after. The shortcomings of this approach are that we do not control for other factors that affect growth and that it is somewhat arbitrary to define what are “major policy reforms”. The advantage is that we can see the annual path of growth rates and thus get a better test of the different prediction for post-reform transitional dynamics made by the models in the theoretical section. One ambitious attempt to identify major policy reform episodes was made by Sachs and Warner (1995). Sachs and Warner rate an economy as closed if any of the following hold: (1) a black market premium more than 20 percent, (2) the government has a purchasing monopoly at below-market prices on a major commodity export, (3) the country has a socialist economic system, (4) non-tariff barriers cover more than 40 percent of intermediate and capital goods imports, and (5) weighted average tariff of more than 40 percent on intermediate and capital goods. Note that only some of these criteria have anything to do with “trade openness” in the usual sense, as pointed out by Rodriguez and Rodrik (2001). The important thing for my purposes is that Sachs and Warner identify the dates of “reform” according to these criteria. I utilize an updated series of Sachs– Warner openness that goes through 1998.19 I pick out countries with at least 13 years of growth data after opening. Since most openings happen towards the end of the sample period, this limits the sample of countries to only 13: Botswana, Chile, Colombia, Costa Rica, Ghana, Guinea, the Gambia, Guinea-Bissau, Israel, Mexico, Morocco, New Zealand, and Papua New Guinea. Figure 16 shows the path of growth and investment before and after opening, after first smoothing each country’s series individually with an HP filter. The results do not support any of the above policies and growth models very convincingly. Investment is completely at variance with the predictions for its transitional path. Growth does show a steady acceleration after opening. This could be either a symptom of increasing returns or simply a process of increased credibility as the reforms take hold. Note however that growth was highest many years before the opening. Perhaps the story of closed and open economies is something more complex like temporary high growth under import substitution, which eventually crashed, followed by an opening of the economy and a partial recovery of growth.
19 The source is Easterly, Levine and Roodman (2004).
Ch. 15:
National Policies and Economic Growth
1051
Figure 16. Growth and investment before and after opening economy in 13 countries.
One of the few cases to fit the predictions of growth models as to transitional dynamics is Ghana, where both investment and growth increase after opening. Both keep rising after the date of opening, again supporting either an increasing returns story or increasing credibility of reform (see Figure 17). Another type of reform that lends itself to transition analysis is stabilization from high inflation. I record episodes of high inflation as following the above definition (log
1052
W. Easterly
Figure 17. Growth and investment before and after opening economy in Ghana.
rate of inflation above 0.3). I measure years of high inflation prior to stabilization, and then years after stabilization when inflation remains below 0.3. I require that there be at least two years of high inflation to rule out one-time spikes in the price level. The first year after inflation comes down is recorded as year 1. Figure 18 shows the behavior of growth and investment before and after inflation comes down. Growth fits the prediction of theoretical models in jumping to a higher path immediately after inflation comes
Ch. 15:
National Policies and Economic Growth
1053
Figure 18. Investment and growth after inflation stabilization.
down. We only have a large enough sample for 7 years after inflation comes down, but growth seems to remain fairly constant post-stabilization. Investment fails to fit the transition predictions of any of the models. We are left with a somewhat mixed picture. There is a fairly rapid growth effect after policy reform, either accelerating or constant. Investment in physical capital does not seem to respond to reforms in the way predicted by growth models. Of course,
1054
W. Easterly
causality is up for grabs. There is also still the extreme policies problem, as episodes in which the country was closed or inflation was very high reflect asymmetrically destructive policies; it is not surprising that growth rebounds after these policies are terminated.
7. Institutions versus policies Recent research has examined the relative role of historical institutions and more recent government policy behavior. The institutions view holds that geographic and historical conditions produce long-lasting differences in institutions. For example, environments where crops are most effectively produced using large plantations will quickly develop political and legal institutions that protect the few landholders from the many peasants [Engerman and Sokoloff (1997)]. Even when agriculture recedes from the economic spotlight, enduring institutions will continue to thwart competition and hence economic development. Similarly, many countries’ institutions were shaped during colonization, so that examining colonies is a natural experiment. European colonialists found different disease environments around the globe. In colonies with inhospitable germs and climates, the colonial powers established extractive institutions, so that a few colonialists could exploit natural resources. In colonies with hospitable climates and germs, colonial powers established settler institutions. According to this view, the institutional structures created by the colonialists in response to the environment endure even with the end of colonialism [Acemoglu, Johnson and Robinson (2001, 2002)]. A history of ethnolinguistic divisions may both prevent the development of good institutions and be more damaging when those institutions are absent [Mauro (1995), Easterly and Levine (1997), and Easterly (2001)]. Thus, the institution view argues that economic development mainly depends on institutions that reflect deep-seated historical factors [North (1990)]. In contrast, the policy view – which is really a collection of many different approaches – questions the importance of history or geography in shaping economic development today. This view is embedded in the approach of multilateral development institutions. The policy view holds that economic policies and institutions reflect current knowledge and political forces. Thus, changes in either knowledge about which policies and institutions are best for development or changes in political incentives will produce rapid changes in institutions and economic policies. According to the policy view, while history and geography may have influenced production and institutions, understanding them is not crucial to understanding economic development today. Easterly and Levine (2003) examine whether major macroeconomic policies – inflation, trade policies, and impediments to international transactions as reflected in real exchange rate overvaluation – help explain current levels of economic development, after controlling for institutions. They do this in two steps. First, they treat the macroeconomic policy indicators, which are averaged over the last four decades as exogenous.
Ch. 15:
National Policies and Economic Growth
1055
Simultaneity bias may bias these results toward finding a significant statistical relationship between policies and economic development if economic success tends to produce better policies. Second, they treat the macroeconomic policy indicators as endogenous; they use instrumental variables (geographic variables and ethnolinguistic fractionalization) to control for potential simultaneity bias. Using these two methods, they assess whether macroeconomic policies explain cross-country differences in economic development. In both methods they instrument for institutions with the set of variables discussed above. The evidence suggests that macroeconomic policies do not have a significant impact on economic development after accounting for the impact of institutions on the level of economic development. When the policy variables are treated as included exogenous variables, the Institutions Index enters all of the regression significantly. Furthermore, the coefficient size on the Institutions Index is essentially unchanged from regressions that did not include policy indicators. Thus, even after controlling for macroeconomic policies, institutions explain cross-country differences in economic development. Furthermore, the data never reject the OIR-test. The policy indicators never enter the regressions significantly. Inflation, Openness, and Real Exchange Rate Overvaluation never enter with a P -value below 0.10. Moreover, even when they are included together, the data do not reject the null hypothesis that the three policies all enter with coefficients equal to zero, which is shown using the F-test on the three policy variables. When using instrumental variables for the policy indicators, they again find that macroeconomic policies do not explain economic development. Specifically, they fail to reject that hypothesis that macroeconomic policies have zero impact on economic development after accounting for the impact of institutions. As noted earlier, the instrumental variables explain a significant amount of the crosscountry variation in the Institutions Index. In the first-stage regressions for policy, Easterly and Levine (2003) find that the instruments explain a significant amount of the cross-country variation in Openness and Real Exchange Rate Overvaluation at the 0.01 significance level. However, the instruments do not do a very good job of explaining cross-country variation in inflation, i.e., they fail to find evidence that the instruments explain average inflation rates over the last four decades at the 0.01 significance level. The policy variables never enter significantly in either method. While the exogenous component of the Institutions Index (i.e., the component defined by endowments) continues to significantly account for international differences in the level of GDP per capita, the macroeconomic policy indicators do not add any additional explanatory power. This raises the suspicion that adverse macroeconomic policies (and macroeconomic volatility in general) may have been proxying for poor institutions in growth regressions. Acemoglu et al. (2003) provide some evidence supporting this suspicion. In sum, the long run effect of policies on development is difficult to discern once you also control for institutions.
1056
W. Easterly
8. Conclusions The large literature on national policies and growth established some statistical association between national economic policies and growth. I confirm that association in this paper and I show how it could have reasonable theoretical foundations. However, I find that the associations seem to depend on extreme values of the policy variables, that the results are not very robust to different econometric methods or introducing initial income, and that a levels regression does not show any effect of policies after controlling for institutions (both instrumented for possible endogeneity). These results are consistent with other theoretical models that predict only modest effects of national policies, depending on model parameters, and show nonlinear effects of tax-cum-subsidy schemes. They are also consistent with the view that the residual A explains most of income and growth differences, and it likely reflects deep-seated institutions that are not very amenable to change in the short run.
References Acemoglu, D., Johnson, S., Robinson, J. (2001). “The colonial origins of comparative development”. American Economic Review. Acemoglu, D., Johnson, S., Robinson, J. (2002). “Reversal of fortunes: Geography and institutions in the making of the modern world income distribution”. Quarterly Journal of Economics 117 (in press). Acemoglu, D., Johnson, S., Robinson, J., Thaicharoen, Y. (2003). “Institutional causes, macroeconomic symptoms: Volatility, crises and growth”. Journal of Monetary Economics (in press). Ades, A., Glaeser, E. (1999). “Evidence on growth, increasing returns, and the extent of the market”. Quarterly Journal of Economics CXIV (3), 1025–1046. Aghion, P., Howitt, P. (1998). Endogenous Growth Theory. MIT Press. Arellano, M., Bond, S. (1991). “Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations”. Review of Economic Studies 58 (2), 277–297. Azariadis, C., Drazen, A. (1990). “Threshold externalities in economic development”. Quarterly Journal of Economics 105, 501–526. Barro, R.J. (1991). “Economic growth in a cross section of countries”. Quarterly Journal of Economics 106 (2), 407–443. Barro, R.J. (1995). “Inflation and economic growth”. Bank of England Quarterly Bulletin, May, 166–176. Barro, R.J. (1998). Determinants of Economic Growth: A Cross Country Empirical Study. MIT Press, Cambridge, MA. Barro, R.J., Lee, J.-W. (2000). “International data on educational attainment: Updates and implications”, Working Paper No. 42. Center for International Development, Harvard University. Barro, R., Sala-i-Martin, X. (1992). “Convergence”. Journal of Political Economy. Becker, G.S., Murphy, K.M., Tamura, R. (1990). “Human capital, fertility, and economic growth”. Journal of Political Economy 98 (5), S12–S37. Part 2. Bénabou, R. (1993). “Workings of a city: Location, education, and production”. Quarterly Journal of Economics 108 (August), 619–652. Bénabou, R. (1996). “Heterogeneity, stratification, and growth: Macroeconomic implications of community structure and school finance”. American Economic Review 86 (3), 584–609. Benhabib, J., Spiegel, M. (1994). “Role of human capital in economic development: Evidence from aggregate cross-country data”. Journal of Monetary Economics 34, 143–173.
Ch. 15:
National Policies and Economic Growth
1057
Borjas, G.J. (1992). “Ethnic capital and intergenerational mobility”. Quarterly Journal of Economics 107 (February), 123–150. Borjas, G.J. (1995). “Ethnicity, neighborhoods, and human capital externalities”. American Economic Review 85 (3), 365–390. Borjas, G.J. (1999). Heaven’s Door: Immigration Policy and the American Economy. Princeton University Press, Princeton. Bouillon, C.P., Legovini, A., Lustig, N. (2003). “Rising inequality in Mexico: Household characteristics and regional effects”. Journal of Development Studies 39 (4), 112–133. Bruno, M., Easterly, W. (1998). “Inflation crises and long-run growth”. Journal of Monetary Economics 41, 3–26. Caselli, F., Esquivel, G., Lefort, F. (1996). “Reopening the convergence debate: A new look at cross-country growth empirics”. Journal of Economic Growth 1 (3), 363–389. Chari, V.V., Kehoe, P.J., McGrattan, E.R. (1996). “The poverty of nations: A quantitative exploration”. Staff Report 204, Federal Reserve Bank of Minneapolis. Collier, P., Dollar, D. (2001). “Can the world cut poverty in half? How policy reform and international aid can meet international development goals”. World Development. Dollar, D. (1992). “Outward-oriented developing economies really do grow more rapidly: Evidence from 95 LDCs, 1976–1985”. Economic Development and Cultural Change 40 (3), 523–544. Easterly, W. (1993). “How much do distortions affect growth?”. Journal of Monetary Economics 32 (November), 187–212. Easterly, W. (1994). “Economic stagnation, fixed factors, and policy thresholds”. Journal of Monetary Economics 33, 525–557. Easterly, W. (2001). The Elusive Quest for Growth: Economists’ Adventures and Misadventures in the Tropics. MIT Press, Cambridge, MA. Paperback edition 2002. Easterly, W., Levine, R. (1997). “Africa’s growth tragedy: Policies and ethnic divisions”. Quarterly Journal of Economics, November. Easterly, W., Levine, R. (2001). “It’s not factor accumulation: Stylized facts and growth models”. World Bank Economic Review. Easterly, W., Levine, R. (2003). “Tropics, germs, and crops: The role of endowments in economic development”. Journal of Monetary Economics 50 (1). Easterly, W., Levine, R., Roodman, D. (2004). “Aid, policies, and growth: Comment”. American Economic Review 94 (3), 774–780. Easterly, W., Rebelo, S. (1993a). “Fiscal policy and economic growth: An empirical investigation”. Journal of Monetary Economics 32, 417–458. Easterly, W., Rebelo, S. (1993b). “Marginal income tax rates and economic growth in developing countries”. European Economic Review 37, 409–417. Easterly, W., Kremer, M., Pritchett, L., Summers, L. (1993). “Good policy or good luck? Country growth performance and temporary shocks”. Journal of Monetary Economics 32 (December), 459–483. Engerman, S., Sokoloff, K. (1997). “Factor endowments, institutions, and differential paths of growth among new world economies: A view from economic historians of the United States”. In: Haber, S. (Ed.), How Latin America Fell behind. Stanford University Press, Stanford, CA. Fischer, S. (1993). “The role of macroeconomic factors in growth”. Journal of Monetary Economics 32, 485– 512. Frankel, J.A., Romer, D. (1999). “Does trade cause growth?”. American Economic Review 89, 379–399. Fujita, M., Krugman, P., Venables, A. (Eds.) (1999). The Spatial Economy: Cities, Regions, and International Trade. MIT Press, Cambridge, MA. Glaeser, E., Laibson, D., Sacerdote, B. (2002). “An economic approach to social capital”. Economic Journal 112, 437–458. Gylfason, T. (1998). “Output gains from economic stabilization”. Journal of Development Economics 56 (1), 81–96. Hall, R.E., Jones, C. (1999). “Why do some countries produce so much more output per worker than others?”. Quarterly Journal of Economics 114 (1), 83–116.
1058
W. Easterly
Hanushek, E.A., Kimko, D.D. (2000). “Schooling, labor-force quality, and the growth of nations”. American Economic Review 90 (5), 1184–1208. Hsieh, C.-T., Klenow, P. (2005). “Relative prices and relative prosperity”. Mimeo, Stanford University. International Monetary Fund (2000). Policies for Faster Growth and Poverty Reduction in Sub-Saharan Africa and the Role of the IMF. Issues Brief. Washington, DC. Jalan, J., Ravallion, M. (1997). “Spatial poverty traps?”. World Bank Policy Research Working Paper No. 1862. Kaufmann, D., Kraay, A., Zoido-Lobatón, P. (1999). “Governance matters”. World Bank Research Working Paper 2196. Kenny, C., Williams, D. (2001). “What do we know about economic growth? Or, why don’t we know very much?”. World Development 29 (1), 1–22. Klenow, P. (1998). “Ideas versus rival human capital: Industry evidence on growth models”. Journal of Monetary Economics 42, 2–23. Klenow, P., Rodriguez-Clare, A. (1997a). “Economic growth: A review essay”. Journal of Monetary Economics 40, 597–617. Klenow, P., Rodriguez-Clare, A. (1997b). “The neoclassical revival in growth economics: Has it gone too far?”. NBER Macroeconomics Annual 1997 12, 73–103. Klitgaard, R., Fitschen, A. (1997). “Exploring income variations across traditional authorities in KwaZuluNatal, South Africa”. Development Southern Africa 14 (3). Kosmin, B.A., Lachman, S.P. (1993). One Nation under God: Religion in Contemporary American Society. Harmony Books. Kremer, M. (1993). “O-ring theory of economic development”. Quarterly Journal of Economics 108 (August), 551–575. Krueger, A.B., Lindahl, M. (2001). “Education for growth: Why and for whom?”. Journal of Economic Literature 39 (4), 1101–1136. Krugman, P.R. (1991). Geography and Trade. MIT Press, Cambridge, MA. Krugman, P. (1995). Development, Geography, and Economic Theory. MIT Press, Cambridge, MA. Krugman, P. (1998). “Space: The final frontier”. Journal of Economic Perspectives 12 (Spring), 161–174. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W. (1998). “Law and finance”. Journal of Political Economy 106, 1113–1155. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W. (1999). “The quality of government”. Journal of Law, Economics, and Organization 15, 222–279. Levine, R. (2005). “Law, endowments, and property rights”. Journal of Economic Perspectives 19 (3), 61–88. Levine, R., Loayza, N., Beck, T. (2000). “Financial intermediation and growth: Causality and causes”. Journal of Monetary Economics 46 (August), 31–77. Levine, R., Renelt, D. (1992). “A sensitivity analysis of cross-country growth regressions”. American Economic Review 82 (4), 942–963. Lipset, S.M. (1996). American Exceptionalism: A Double-Edged Sword. W.W. Norton, New York. Mankiw, N.G. (1995). “The growth of nations”. Brookings Papers on Economic Activity 1, 275–326. Mankiw, N.G., Romer, D., Weil, D.N. (1992). “Contribution to the empirics of economic growth”. Quarterly Journal of Economics 107, 407–437. McGrattan, E., Schmitz, J. (1998). “Explaining cross-country income differences”. Mimeo, June. Federal Reserve Bank of Minneapolis, Research Department. Murphy, K.M., Shleifer, A., Vishny, R. (1989). “Industrialization and the big push”. Journal of Political Economy 97, 1003–1026. North, D.C. (1990). Institutions, Institutional Change and Economic Performance. Cambridge University Press, Cambridge, UK. Patrinos, H.A. (1997). “Differences in education and earnings across ethnic groups in Guatemala”. Quarterly Review of Economics and Finance 37 (4). Pritchett, L. (1997). “Where has all the education gone?”. World Bank Policy Research Working Paper No. 1581.
Ch. 15:
National Policies and Economic Growth
1059
Psacharopoulos, G., Patrinos, H.A. (1994). “Indigenous people and poverty in Latin America”. Human Resources Development and Operations Policy Working Paper No. 22. Putnam, R.D. (1993). Making Democracy Work: Civic Traditions in Modern Italy. With R. Leonardi and R.Y. Nanetti. Princeton University Press, Princeton, NJ. Putnam, R.D. (2000). Bowling Alone: The Collapse and Revival of American Community. Simon & Schuster, New York. Ravallion, M., Jalan, J. (1996). “Growth divergence due to spatial externalities”. Economics Letters 53, 227– 232. Ravallion, M., Wodon, Q. (1998). “Poor areas or only poor people?”. Mimeo. World Bank. Rebelo, S. (1991). “Long run policy analysis and long run growth”. Journal of Political Economy 99, 500– 521. Rebelo, S. (1992). “Growth in open economies”. Carnegie–Rochester Conference Series on Public Policy 36 (0), 5–46. Rebelo, S., Stokey, N.L. (1995). “Growth effects of flat-rate taxes”. Journal of Political Economy 103, 519– 550. Rodriguez, F., Rodrik, D. (2001). “Trade policy and economic growth: A skeptic’s guide to the crossnational evidence”. In: NBER Macroeconomics Annual 2000, vol. 15. MIT Press, Cambridge and London, pp. 261–325. Romer, P. (1986). “Increasing returns and long-run growth”. Journal of Political Economy 94, 1002–1037. Romer, P. (1995). “Comment on N. Gregory Mankiw, “The growth of nations””. Brookings Papers on Economic Activity 1, 313–320. Sachs, J., Warner, A. (1995). “Economic reform and the process of global integration”. Brookings Papers on Economic Activity 1, 1–95. Sala-i-Martin, X. (2002). “Fifteen years of new growth economics: What have we learned?”. In: Economic Growth: Sources, Trends, and Cycles. In: Series on Central Banking, Analysis, and Economic Policies, vol. 6. Central Bank of Chile, Santiago, pp. 41–59. Sarel, M. (1996). “Nonlinear effects of inflation on economic growth”. IMF Staff Papers 43 (March), 199– 215. Slemrod, J. (1995). “What do cross-country studies teach about government involvement, prosperity, and economic growth?”. Brookings Papers on Economic Activity 0 (2), 373–415. Summers, R., Heston, A. (1991). “The Penn World Table (Mark 5): An expanded set of international comparisons, 1950–1988”. Quarterly Journal of Economics, 327–368. Temple, J. (1999). “The new growth evidence”. Journal of Economic Literature 37 (1), 112–156. Williamson, J. (1990). “What Washington means by policy reform”. In: Latin American Adjustment: How Much Has Happened?. Institute for International Economics, Washington, DC. Woolcock, M., Narayan, D. (2000). “Social capital: Implications for development theory, research, and policy”. World Bank Research Observer 15 (2), 225–249. World Bank (1981). Accelerated Development in Sub-Saharan Africa. Washington, DC. World Bank (2002). World Development Indicators. World Bank, Washington, DC. Young, A. (1995). “The tyranny of numbers: Confronting the statistical realities of the East Asian growth experience”. Quarterly Journal of Economics 110, 641–680.
AUTHOR INDEX
n indicates citation in a footnote. 1720, 1721, 1721n, 1729n, 1770, 1770n, 1772, 1788, 1800n, 1801n, 1810n Aghion, P., see Acemoglu, D. 71n, 99, 102n, 277n, 428, 871, 871n, 943n, 1000, 1327, 1597n, 1617 Ahluwalia, M.S. 1712, 1732, 1734 Ahluwalia, M.S., see Chenery, H.B. 1714n, 1724n Ahn, S. 633 Ahn, T., see Ostrom, E. 1659 Aiyagari, S.R. 543 Aiyar, C.V., see Timberg, T. 479, 480, 483, 525 Aiyar, S. 684 Aizcorbe, A. 749 Aizenman, J. 38, 654, 655 Akerlof, G. 645, 1655 Albrecht, J. 1338n Alcalá, F. 658, 659, 662, 1096, 1422n, 1516, 1516n, 1519, 1525, 1526, 1530 Aleem, I. 479, 480, 480n, 483 Alesina, A. 278, 398, 399, 538n, 652, 653, 655, 656, 662, 1422n, 1502, 1502n, 1503, 1503n, 1505, 1505n, 1506, 1510, 1510n, 1511n, 1513, 1514n, 1518n, 1519, 1519n, 1525n, 1530, 1530n, 1535n, 1538n, 1597n, 1600n, 1609, 1686, 1688 Allen, F. 871, 876, 881, 882, 884, 886, 909, 918, 918n, 920 Allen, R.C. 436, 1138n Altonji, J.G. 700 Alvarez, F. 41n, 1470n Amable, B. 663 Amir, R. 350 Amsden, A.H. 1001n, 1004n Anand, S. 1713n, 1732 Anant, T.C.A., see Segerstrom, P.S. xiin, 71n, 114, 127n, 1070n, 1425n Anas, A. 1570 Anas, A., see Abdel-Rahman, H. 1565 Anderson, G. 594, 595, 595n Anderson, J.E. 985n
Aaronson, E. 1680n Aaronson, S. 1354n Abdel-Rahman, H. 1564, 1565, 1570, 1574 Ableidinger, J., see Case, A. 518n Aboud, A., see Barrett, C.B. 339, 355 Abowd, J.M. 1331n Abraham, A. 1306n Abramovitz, M. 198, 212, 582, 590, 604, 781, 784, 804, 806 Acemoglu, D. 71n, 99, 102n, 120n, 130, 131, 135, 136, 138n, 140–143, 150, 150n, 154n, 193n, 223n, 277, 277n, 278, 302, 326, 349, 357, 393, 395, 395n, 397n, 398, 402, 410, 413, 414, 417, 417n, 419, 422–424, 428, 430, 432, 435, 438, 448, 450, 453, 458, 461, 463, 464, 500n, 502, 503, 506, 541n, 607, 639n, 651, 654, 656–658, 662, 710n, 734n, 807n, 820, 835, 871, 871n, 876, 880, 883, 923, 938, 943n, 1000, 1005, 1033, 1054, 1055, 1106, 1283, 1298, 1299n, 1305, 1307, 1307n, 1323, 1327, 1328, 1339n, 1351, 1361, 1425n, 1453n, 1472n, 1490n, 1597n, 1598n, 1608n, 1617, 1620n, 1623n, 1625n, 1627, 1720n, 1721, 1741 Addison, J.T. 1330, 1330n Addison, T., see Cornia, G.A. 1733 Adelman, I. 1706n Ades, A.F. 398, 1021, 1422n, 1518, 1560, 1561, 1563, 1564 Adsera, A. 328 Agell, J. 662 Agénor, P.-R. 648 Aghion, P. xiin, 16n, 28, 69, 70, 71n, 82, 82n, 86, 89, 90, 92n, 102, 102n, 104, 106n, 107n, 114, 144, 157, 157n, 223, 256n, 354n, 388, 398, 499, 501, 510n, 529, 537, 539n, 820, 822, 872, 874, 879, 887, 899, 938, 986n, 1018, 1020n, 1069, 1070, 1073, 1087n, 1089, 1090, 1090n, 1092, 1103, 1106, 1116n, 1296, 1302, 1303, 1311, 1313, 1314n, 1315, 1332, 1333, 1347, 1504n, 1517n, 1620n, 1715n, I-1
I-2 Anderson, R.D. 210, 211, 211n Anderson, S. 521n Ando, A., see Modigliani, F. 1425n Andorka, R. 201, 202 Andrade, E. 598 Andreoni, J. 1799 Andres, J. 598, 598n Angeletos, G.-M. 515, 544n Angeletos, G.-M., see Alesina, A. 1597n Angeletos, M., see Aghion, P. 107n, 879 Angrist, J., see Acemoglu, D. 502, 710n Anker, R., see Clark, R. 1725n Annen, K. 1658 Anselin, L. 644 Antrás, P. 131n, 135, 1130 Antweiler, A. 1770n, 1771n, 1788 Aoki, M. 981, 983, 1425n Aportela, F. 520 Arellano, M. 541n, 630n, 632, 633, 900, 901, 1048, 1711n Arestis, P. 905 Aristotle, 1641 Armendariz de Aghion, B., see Aghion, P. 107n Arndt, H.W. 1714n Arnold, L.G. 1425n Arrow, K.J. xiv, 78n, 113, 1070, 1070n, 1643n, 1798 Arthur, B. 1551 Arthur, W.B. 302 Artzrouni, M., see Komlos, J. 223n Asbell, B. 1725n Ashenfelter, O.C. 1727n Ashraf, N. 522 Aslund, A. 989, 999 Aston, T.H. 441 Aten, B., see Heston, A. xii, 303n, 491, 562, 685, 804n, 826, 828, 830, 831, 834, 1412, 1519 Atje, R. 894 Atkeson, A.A. 1098n, 1204, 1296, 1325, 1360, 1452n Atkinson, A.B. 131n, 1733, 1736n Atkinson, T.R. 1214 Attanasio, O. 1356 Au, C.C. 1585 Auerbach, A.J. 847n Autor, D.H. 1283n, 1302n, 1597n Autor, D.H., see Katz, L.F. 1281, 1281n, 1282, 1285, 1715n, 1720n, 1724, 1727n
Author Index Azariadis, C. 164n, 256n, 303, 317, 318n, 331, 333, 373, 398, 588, 597, 607, 653, 999, 1000, 1021, 1681 Bacha, E.L. 1714 Backus, D.K. 1096, 1425n, 1517 Bagehot, W. 867, 871, 880 Bahadur, C., see Sachs, J.D. 298, 300n, 304n Bahk, B.H. 1204, 1294, 1295n, 1296 Bailey, R.W., see Addison, J.T. 1330, 1330n Baily, M.N. 718, 771n, 779n Bairoch, P. 185, 192, 193, 282, 282n, 410, 1373, 1373n, 1377, 1377n, 1532 Baker, E. 316, 349, 368 Baker, G.P. 1322, 1325 Baland, J.-M. 1608n Baland, J.-M., see Anderson, S. 521n Baldwin, J.R. 786n Baldwin, R.E. 281n, 1425n, 1486n, 1579, 1724 Baltagi, B. 644 Bandiera, O. 339, 349 Bandyopadhyay, S. 598, 598n Bandyopadhyay, S., see Shafik, N. 1766n Banerjee, A.V. 92n, 354n, 355n, 356, 398, 399, 441, 479, 480n, 481, 483, 494, 495, 507–511, 511n, 513, 515–518, 525, 528, 530, 537, 539n, 541, 544n, 617, 887, 1604n, 1605n, 1623, 1720, 1722, 1733n Banerjee, A.V., see Aghion, P. 82n, 107n, 879 Banfield, E.C. 402 Banks, R.B. 939–941 Baqir, R., see Alesina, A. 1505n Barbier, E. 1810, 1814n Bardham, P. 1425n Bardhan, P. 364n, 367, 398, 501n, 513, 1505n Barlevy, G. 33, 34, 45 Barnett, H.J. 1751n Barr, A. 1653, 1657 Barret, S. 1798n Barrett, C.B. 326n, 339, 355 Barro, R.J. xi, xin, 16n, 25, 28, 77, 78n, 105, 120n, 122n, 139n, 195, 196n, 216, 224n, 283n, 402, 403, 491n, 495, 518, 581, 582n, 586n, 587, 587n, 589, 591, 592, 540n, 541, 604, 604n, 605, 606, 618, 625, 630, 638, 647, 651–660, 662, 685, 694, 695, 821, 826, 827, 829–831, 834, 835, 837–839, 846, 938, 940, 944, 945n, 948, 1037, 1046, 1044n, 1046, 1099n, 1103n, 1445, 1509n, 1519, 1523, 1560, 1612, 1733n Barro, R.J., see Alesina, A. 1505n Barro, R.J., see Becker, G.S. 1261, 1262
Author Index Barro, R.J., see Lee, J.-W. 701, 702, 703n, 704 Barro, R.M. 24 Bartel, A.P. 938, 1205, 1301, 1318 Bartels, L. 1608 Barth, J.R. 922 Barzel, Y. 436 Basta, S.S. 490 Basu, A. 283n Basu, K. 373 Basu, S. 734n, 938, 942, 943, 1296, 1427n Bates, R.H. 443–445 Batou, J., see Bairoch, P. 410 Battalio, R.C., see Van Huyck, J.B. 330 Battistin, E., see Attanasio, O. 1356 Baumol, W.J. 77n, 582, 586n, 590, 604, 786, 804, 827, 1117, 1715n Baxter, M. 1425n Bayart, J.-F. 1671n Beath, J. 349 Beaudry, P. 660, 1304 Beck, T. 653, 655, 661, 897, 897n, 900, 902, 904, 904n, 907n, 912, 913n, 914–920, 922, 923, 923n Beck, T., see Levine, R. 84, 661, 897, 898, 898n, 899, 900, 1033 Becker, C., see Mills, E. 1560 Becker, G.S. 224n, 227, 228, 229n, 235, 423, 518, 780n, 852n, 1021, 1261–1264, 1300, 1378, 1578 Becker, G.S., see Barro, R.J. 224n Becker, R., see Henderson, J.V. 1568, 1574 Becker, X. 1603n Beeson, P.E. 1558 Behrens, W.W., see Meadows, D.H. 1752n Behrman, J.R. 709 Bekaert, G. 653, 661, 907 Bekar, C., see Lipsey, R.G. 1295 Bell, B., see Nickell, S. 1350 Bell, C.L.G., see Chenery, H.B. 1714n, 1724n Ben Porath, Y. 258n Ben-David, D. 659, 1091n, 1514 Ben-Zion, U., see Razin, A. 224n, 1261, 1264 Bénabou, R. 259n, 538n, 541, 1021, 1574, 1597n, 1599, 1600, 1602n–1605n, 1606, 1607n, 1609, 1610n, 1618n, 1619, 1630, 1715n, 1733n Bénabou, R., see Fukao, K. 1485n Benassy, J.-P. 120 Bencivenga, V.R. 150n, 875, 877, 878, 895, 896, 923
I-3 Benhabib, J. 81, 398, 495, 538n, 903, 938, 939, 953, 959, 959n, 1018n, 1263 Bennell, P. 484, 484n, 487 Bentolila, S. 1340 Berdugo, B. 279n Berdugo, B., see Hazan, M. 230, 230n, 236n, 260–262, 1266 Berg, M. 1160n Berger, A.N. 897 Bergh, A.E., see Lipscomb, A.A. 1069 Berghahn, V.R. 214 Bergsman, J. 1555 Berkowitz, D. 1007, 1008 Berle, A.A. 873 Berman, B.J. 449 Berman, E. 1597n Bernanke, B.S. 82n, 731, 731n Bernard, A. 584, 588, 599–602, 604n Berndt, E.R. 1196n Bernhardt, D., see Lloyd-Ellis, H. 537n Bernstein, L. 1647, 1650 Berry, T.S. 1218n, 1220n Berthelemy, J. 653 Bertocchi, G. 193n, 259n, 263, 264, 278n Bertola, G. 1312, 1336n, 1340, 1342, 1715n, 1718n Bertrand, M. 522, 909 Besley, T.F. 429n, 508, 516, 650, 999, 1005 Bessen, J. 145, 1204 Beugelsdijk, S. 658, 1682 Bezuneh, M., see Barrett, C.B. 339, 355 Bhargava, A. 655 Bhattacharya, S. 871 Bhide, A. 882, 895 Bianchi, M. 331, 594, 621n Bianchi, S. 1685, 1686 Bigsten, A. 481, 1650 Bikhchandani, S. 516n Bils, M. 495, 502, 653, 687n, 698, 706, 833, 938, 1104n Binder, M. 581, 592, 627, 645 Binswanger, H. 508 Binswanger, H., see Rosenzweig, M.R. 515 Birdsall, N. 974n, 1725n Bisin, A. 266n Black, D. 1548, 1550, 1553–1555, 1558, 1564, 1569, 1571, 1574, 1577 Black, S.W. 884 Blackbourn, D. 1649n Blackburn, K. 871 Blades, D. 784n
I-4 Blanchard, O. 1288, 1340, 1350, 1351 Blanchflower, D. 511n, 1692n Blattman, C. 656 Blau, F.D. 1266, 1267, 1267n, 1731 Blau, F.D., see Bertola, G. 1312, 1340 Blaut, J.M. 1170n Bleaney, M. 1327 Blinder, A.S., see Baumol, W.J. 1715n Bliss, C. 593 Blomstrom, M. 627, 656, 659 Blonigen, B. 654 Bloom, D.E. 256, 261n, 277n, 335, 400, 621, 652, 654, 655, 657 Bloom, B., see Aghion, P. 89, 90, 102n Bloom, N., see Aghion, P. 1504n Blum, J. 433 Blume, L. 1656 Blundell, R. 76, 86, 89, 633, 1331n, 1355, 1356, 1711n Blundell, R., see Aghion, P. 89, 90, 102n, 1504n Board of Governors of the Federal Reserve System, 1213n Bockstette, V. 662 Bodenhorn, H. 923n Bodie, Z., see Merton, R.C. 869, 886, 919 Boisjoly, J., see Hofferth, S. 1687 Boldrin, M. 28, 30, 62, 234n, 1088, 1207, 1265 Bolton, P. 1320, 1321, 1504n, 1506, 1510n Bolton, P., see Aghion, P. 92n, 354n, 537, 539n, 887, 1720 Bond, E. 26 Bond, S. 627, 632, 660 Bond, S., see Arellano, M. 541n, 632, 900, 1048, 1711n Bond, S., see Blundell, R. 633, 1711n Boone, J. 89 Boone, P., see Aslund, A. 989 Boot, A.W.A. 881–883 Booth, A. 1327 Borghans, J.A.M. 266 Borghans, L. 1303 Borghans, L., see Borghans, J.A.M. 266 Borjas, G.J. 142n, 1021, 1035 Boserup, E. 239, 239n, 1069 Bosworth, B.P. 660, 746n, 771n, 972 Bosworth, B.P., see Collins, S.M. 996n Botticini, M. 1155 Bottomley, A. 480n Boucekkine, R. 259 Bouillon, C.P. 1035
Author Index Bound, J., see Berman, E. 1597n Bourguignon, F. 259n, 399, 519, 1441, 1608n, 1714, 1717n, 1719n, 1730n, 1734n, 1735, 1737n, 1739n, 1741n Bourguignon, F., see Levy-Leboyer, M. 188, 214 Boustan, L., see Aghion, P. 102n Bovenberg, A.L. 1778, 1797 Bover, O., see Arellano, M. 633, 901 Bowles, S. 212n, 259n, 266n, 303, 1643, 1650n Bowlus, A.J. 1354, 1356, 1357 Boyce, S., see Gertler, P.J. 498 Boyce, W.E. 1099n Boyd, J.H. 871, 874, 875, 879, 886, 920, 923 Boyd, R. 266n Boyer, G. 272n Braden, J.B., see Chimeli, A.B. 1810n Bradstreet Co., 1200n Brady, D.S. 1196n Brady, H. 1741n Brander, J.A. 1752n Brandolini, A., see Atkinson, A.B. 1733, 1736n Branstetter, L.G. 938 Braudel, F. 150, 1530n Braverman, L. 1311 Brehm, J. 1685, 1686 Breiman, L. 619n Brems, H. 1425n Brennan, R., see Lochner, K. 1683 Brenner, R. 427, 440, 441, 455 Bresnahan, T.F. 1185, 1295, 1322 Breusch, T. 644 Brewer, J. 456 Brezis, E.S. 278n Brezis, S.L. 1427n, 1453n Bridgman, B. 1414 Brinton, M., see Lee, S.J. 1673, 1675 Britnell, R.H. 1118n Brock, S., see Binder, M. 627 Brock, W.A. 16, 308n, 320, 559n, 560n, 581, 609, 610, 611n, 612–615, 639, 645, 971n, 1664, 1665, 1667, 1668, 1670, 1681, 1684, 1692, 1756, 1770n, 1772, 1772n, 1799, 1800, 1809, 1811n, 1815, 1816 Brock, W.H. 1138n, 1152 Broner, F. 1490n Brown, G.I. 1141 Browning, M. 235n, 1739n Bruland, T. 1169
Author Index Bruno, M. 655, 1044n, 1048n Brunton, D. 455 Bryant, L. 1140, 1140n Brynjolfsson, E. 806n, 807n, 1322n Brynjolfsson, E., see Bresnahan, T.F. 1322 Buchanan, R.A. 1164 Buera, F. 537, 543 Buiter, W.H. 1425n Buka, S., see Lochner, K. 1683 Bulli, S. 596, 597 Bun, M. 634 Bureau of Economic Analysis, 751, 754, 784 Bureau of Labor Statistics, 759n, 784, 806 Burgess, R., see Aghion, P. 986n Burgess, R., see Besley, T.F. 650, 999, 1005 Burt, R. 1660 Burton, A. 1139 Byushgens, S.S., see Konus, A. 750 Caballero, R.J. 338n, 1204, 1350 Cagetti, M. 542 Cai, F., see Lin, J.Y. 980 Cain, G.G. 1266 Cain, G.G., see Baldwin, R.E. 1724 Cain, L. 1163n, 1311 Calderón, C. 662 Calderón, C., see Loayza, N.V. 978n Caldwell, J.C. 1265 Caldwell, W.J. 234n Calomiris, C. 878n Calvet, L., see Angeletos, G.-M. 515, 544n Calvo, G. 986n Cameron, D. 1741 Cameron, R. 208n, 453, 909, 910, 910n Campbell, J.Y. 768n, 806n, 1342 Campos, N. 627 Canning, D., see Bloom, D.E. 256, 277n, 335, 621, 654, 655 Cannon, E. 593, 1103 Canova, F. 618, 619, 622 Canova, F., see Bertocchi, G. 193n, 278n Cantor, N.F. 440 Caplow, T. 1251 Caprio Jr., G. 874, 885, 923n Caprio Jr., G., see Barth, J.R. 922 Carbonaro, W. 1680n Card, D. 694, 702n, 1285n, 1313, 1330, 1330n, 1360, 1727n Card, D., see Ashenfelter, O.C. 1727n Cardosso, F.H. 428 Cardwell, D.S.L. 1128n, 1150
I-5 Carlaw, K., see Lipsey, R.G. 1295 Carlin, W. 920 Carlino, G. 602 Caroli, E. 1322, 1619n Caroli, E., see Aghion, P. 1715n, 1721, 1721n Carosso, V. 879 Carpenter, S.R., see Scheffer, M. 1760n Carroll, C. 543 Carson, R.T. 1813 Carter, M. 1674, 1689 Cartwright, F.F. 1169n Carvalho, I. 519 Carver, K., see Teachman, J. 1676, 1679 Case, A. 518n Case, A., see Besley, T.F. 516 Casella, A. 1510n Casella, A., see Rauch, J.E. 1653 Caselli, F. 136n, 140, 277n, 499, 507, 535, 543, 579, 589, 607, 629, 632, 638, 653, 654, 656, 683, 690, 711, 711n, 712n, 714, 714n, 723n, 724, 725, 725n, 728, 733, 734, 734n, 736n, 953, 1048, 1263, 1302, 1302n, 1304, 1315n, 1446n Cashin, P. 586n Cass, D. xi, 16, 17, 298, 388 Castilla, E., see Fernandez, R. 1678 Castillo, M., see Carter, M. 1689 Castro, R., see Cohen-Pirani, D. 1301n Cavalcanti-Ferreira, P., see Pessoa, S. 483 Cavalli-Sforza, L.L. 266n Caves, R.E. 842 Ceballos, M., see Palloni, A. 1673, 1675, 1680 Central Intelligence Agency, 1519 Cervellati, M. 259, 1145 Cetorelli, N. 156, 913 Chakraborty, S. 882 Chamberlain, G. 805 Chanda, A. 279n, 727n Chanda, A., see Bockstette, V. 662 Chandler, T. 410 Chandler Jr., A.D. 755n, 806 Chaney, T. 1342 Chang, R. 1425n Chang, S., see Powell, C. 498 Chari, V.V. 1019n Charkham, J. 885 Charles, K. 1686 Charlesworth, A. 436 Chaudhuri, K.N. 282 Chaudhuri, S., see Ravallion, M. 513 Checchi, D. 356
I-6 Checkland, S.G. 910n Chen, D.L., see Kremer, M. 224n, 272n Chen, S. 297n Chen, S., see Ravallion, M. 1736n Chen, X., see Li, R. 490 Chenery, H.B. 1712, 1714n, 1724n Chesher, A. 618 Chesnais, J. 202, 226n Chester, L.A., see Birdsall, N. 1725n Chèvre, P., see Bairoch, P. 410 Chiappori, P.-A., see Bourguignon, F. 519 Chimeli, A.B. 1810n Chitnis, A. 1134n, 1163n Cho, D. 660 Choi, S.-Y. 751n, 807n Chong, A., see Calderón, C. 662 Chow, G.C. 751 Christensen, C.M. 807n Christensen, K., see Kohler, H. 269n Christensen, K., see Rodgers, J.L. 269n Christensen, L.R. 684n, 781n, 787, 805 Christian, M.S. 54n Christopher, H., see Aghion, P. 938 Christopoulos, D.K. 906 Chui, M. 1425n Chun, H. 1303 Ciccone, A. 145, 145n, 656 Ciccone, A., see Alcalá, F. 658, 659, 662, 1096, 1422n, 1516, 1516n, 1519, 1525, 1526, 1530 Ciccone, A., see Matsuyama, K. 349 Cipolla, C.M. 234, 272, 273 Claessens, S. 653, 661, 874, 913, 914 Clark, C. 719n, 1752n, 1760n Clark, G. 181, 182n, 183, 183n, 188, 197, 205, 209, 213, 228, 235, 254, 255, 266, 271, 272, 718n, 1098n, 1099, 1387 Clark, J.S. 1553 Clark, R. 1725n Clarke, G. 921n Clemens, M.A. 282n, 1724n Clow, A. 1150n Clow, N.L., see Clow, A. 1150n Coale, A.J. 190, 200n Coase, R.H. 422, 872n, 1319 Coate, S.T. 513 Coate, S.T., see Besley, T.F. 429n Coatsworth, J.H. 443 Cochrane, E.W. 1144 Cody, M.L. 270n Coe, D.T. 79n, 842, 852n, 938
Author Index Cohen, D. 590, 1354n Cohen, M.N. 275n Cohen, W.M. 78n, 89, 136 Cohen-Pirani, D. 1301n Cole, A.H. 1137n Cole, H. 544 Cole, H., see Ohanian, L.E. 1098n Colecchia, A. 785n, 786n, 1346 Coleman, J. 1642, 1643, 1647, 1677 Coleman, W.J., see Caselli, F. 140, 277n, 607, 711n, 723n, 724, 725, 725n, 728, 733, 734, 734n, 736n, 1263 Collard, F., see Beaudry, P. 660 Collier, P. 567n, 1017 Collier, P., see Bigsten, A. 481, 1650 Collins, S. 996n Collins, S.M., see Bosworth, B.P. 660, 972 Comin, D. 1102n, 1204, 1342 Commission on Macroeconomics, and Health, 490 Congressional Budget Office, 746n Conley, T.G. 339, 349, 516, 645 Connolly, M. 260n Conrad, K. 718n Constant, E.W. 1153 Cook, D. 640, 660 Cook, J.P., see Van Huyck, J.B. 330 Cooley, T.F. 1294 Cooney, E.W. 1168n Cooper, R.W. 145 Copeland, B.R. 1755, 1757n, 1773, 1779, 1786, 1788, 1799n, 1804 Copeland, B.R., see Antweiler, A. 1770n, 1771n, 1788 Cordoba, J.-C. 1576 Cornia, G.A. 1733 Corrado, C. 752 Corrado, L. 601n Corriveau, L. 71n Costa, D. 1678, 1680, 1685, 1686 Coulombe, S. 656 Cowan, R. 1122n Cowen, T. 910n Cowles, A. 1200n Cozzi, G. 1094, 1308 Crafts, N.F.R. 197, 206, 255 Craig, F.W.S. 215 Cressy, D. 272n Crisp, O., see Cameron, R. 909 Croarken, M. 1158n Crouch, T. 1153n
Author Index Crouzet, F. 1157 Cubberly, E.P. 210, 211 Cull, R. 909 Cumings, B. 405n Cummings, D., see Christensen, L.R. 684n, 781n, 787, 805 Cummins, J.G. 1189, 1292, 1293, 1295, 1304, 1305, 1309, 1309n, 1337 Cuñat, A. 1425n, 1463n Currais, L., see Rivera, B. 662 Currie, J. 704, 705n Curtin, P.D. 417 Cutler, D.M. 1355 Da Rin, M. 340, 343 da Silva, L.P., see Bourguignon, F. 1739n Dagenais, D., see Dagenais, M. 642 Dagenais, M. 642 Dahan, M. 224n Dalgaard, C.-J. 165, 224n Dalgaard, C.-J., see Aiyar, S. 684 Dalgaard, C.-J., see Chanda, A. 279n, 727n Darnton, R. 1137, 1157n Darwin, C. 268 Dasgupta, A. 479, 480 Dasgupta, P. 373, 537, 1643n, 1648, 1659, 1688, 1692, 1752n, 1760n Datt, G., see Ravallion, M. 1734n David, P.A. 301, 806n, 1069n, 1132, 1146n, 1184, 1187, 1204, 1295 David, P.A., see Abramovitz, M. 198, 212, 806 Davids, K. 1137n, 1166 Davidson, R. 643 Davis, D. 1446n Davis, J. 1560, 1561, 1564, 1575 Davis, R. 453 Davis, S.J., see Attanasio, O. 1356 Dawkins, R. 269 Dawson, J.W. 37 de Barros, R.P., see Ferreira, F. 1737n de Figueiredo, R. 1505n de Gregorio, J. 654, 878 de Hek, P.A. 39, 54n, 61, 61n de Janvry, A. 1714 de la Croix, D. 224n de la Croix, D., see Boucekkine, R. 259 de la Fuente, A. 694, 875 de la Torre, A., see Birdsall, N. 974n de Mello, L. 1609 de Menil, G. 1008 de Nardi, M., see Cagetti, M. 542
I-7 de Soto, H. 364 de Vries, J. 150, 185, 193, 453, 454, 1161n Dean, E.R. 784n DeAngelo, H. 882 Dear, P. 1135n Deardorff, A.V. 1453n, 1463n Deaton, A. 513, 543, 544n Debreu, G. 19 Dechert, W.D. 350, 1752n Dehejia, R. 907, 908 Deidda, L. 653 Deininger, K. 538, 1732 DeJong, D.N., see Beeson, P.E. 1558 Dellas, H. 54n DeLong, J.B. 573n, 586n, 590, 604, 643, 656, 660, 828, 879, 990, 1065n Demetriades, P.O. 653, 661, 662, 905, 906 Demetriades, P.O., see Arestis, P. 905 Demirgüç-Kunt, A. 881, 886, 889, 894n, 907n, 914, 916–920 Demirgüç-Kunt, A., see Beck, T. 655, 661, 897, 897n, 907n, 914–918, 920, 922, 923, 923n Demsetz, H. 422 Den Haan, W.J. 326, 1329n, 1346 Denison, E.F. xi, 684n, 751, 780, 783n, 784, 803, 803n, 804, 805 Dercon, S. 340, 357, 357n Dercon, S., see Bigsten, A. 1650 Dertouzos, M. 782n Desdoigts, A. 331, 620, 620n, 1681 Deurenberg, P., see Li, R. 490 Devereux, M.B. 124, 126, 126n, 876, 1425n, 1463n Devine, W.D. 1190 Devleeschauwer, A., see Alesina, A. 278, 653, 1505 Dewatripont, M. 884, 920, 989n Dewatripont, M., see Aghion, P. 874 Dewatripont, M., see Bolton, P. 1320, 1321 Diamond, D.W. 874, 875, 877, 878, 878n Diamond, J.M. 277, 400, 413, 857, 1095 Diamond, P. 733n Dickens, R. 1331n, 1356 Dickens, W.T. 1326 Diebold, F. 603 Diewert, W.E. 750, 760n, 783n Dimaria, C.H. 350 DiNardo, J. 1326 DiNardo, J., see Card, D. 1313, 1360
I-8 Dinopoulos, E. 93n, 127n, 142, 1094, 1307, 1308n, 1425n Dinopoulos, E., see Segerstrom, P.S. xiin, 71n, 114, 127n, 1070n, 1425n DiPasquale, D. 1686 DiPrima, R.C., see Boyce, W.E. 1099n Diwan, I. 130 Dixit, A.K. 113, 116, 398, 429n, 1005, 1007, 1008, 1070, 1082, 1446n Djankov, S. 99, 425, 426, 506, 509n, 1006–1008, 1395 Djankov, S., see Claessens, S. 874 Do, T. 508 Dobb, M.H. 427 Dobbs, B.J.T. 1132 Dobkins, L.H. 1548, 1558 Doepke, M. 227, 227n, 230, 236n, 255, 258, 260, 261, 263, 278, 280, 1262, 1262n, 1265, 1266, 1378 Doepke, M., see de la Croix, D. 224n Dollar, D. 652, 657–659, 661, 1033, 1422n, 1425n, 1514, 1734, 1736 Dollar, D., see Collier, P. 1017 Domar, E. 446, 774, 774n, 779 Doménech, R., see de la Fuente, A. 694 Doms, M. 270n, 752, 755, 756 Donovan, A.L. 1138, 1138n Dooley, M. 1319n Doppelhofer, G. 587, 612–615 Dornbusch, R. 143, 1463n, 1468 Dougherty, C. 787, 805 Doughty, D., see Rodgers, J.L. 269n Douglas, P.H. 781 Dow Jones and Company, Inc., 1206n Dowrick, S. 591, 592, 604n, 1409n Draper, D. 559n, 582 Drazen, A., see Azariadis, C. 317, 318n, 331, 588, 597, 607, 653, 999, 1000, 1021, 1681 Dreze, J. 498 Driscoll, J. 644 DuBoff, R.B. 1187n, 1189n, 1191n Duck, N., see Cannon, E. 593 Duffy, J. 257n, 605n, 620, 938 Duflo, E. 482, 495, 502, 516, 519, 521, 522, 694 Duflo, E., see Banerjee, A. 398, 617, 1605n, 1733n Duflo, E., see Banerjee, A.V. 481, 483, 507, 511, 525, 528, 530, 541 Dulberger, E.R. 749, 751, 774n Dulberger, E.R., see Berndt, E.R. 1196n
Author Index Duloy, J.H., see Chenery, H.B. 1714n, 1724n Dun and Bradstreet, Inc., 1210n Duncan, G., see Hofferth, S. 1687 Dunn, T.A., see Altonji, J.G. 700 Dunne, T., see Doms, M. 270n Durand, J. 1725n Duranton, G. 1325n, 1547, 1554, 1557, 1564, 1565, 1574–1576 Durham, W. 266n Durlauf, S.N. 77n, 256, 259n, 318n, 331, 368n, 398, 402, 515, 561n, 579n, 582n, 589, 607, 617–619, 619n, 622, 631, 647, 804n, 942, 1574, 1643n, 1644, 1664, 1667, 1667n, 1668–1671, 1680n, 1681, 1690, 1693n, 1810 Durlauf, S.N., see Bernard, A. 584, 588, 599–602 Durlauf, S.N., see Bowles, S. 303 Durlauf, S.N., see Brock, W.A. 559n, 560n, 581, 609, 610, 611n, 612–615, 639, 645, 971n, 1664, 1665, 1667, 1668, 1670, 1681, 1684, 1692 Dybvig, P.H., see Diamond, D.W. 877 Dyck, A. 917, 918n Dyson, T. 190n Eamon, W. 1132 Earls, F., see Sampson, R. 1687, 1688 Easterlin, R.A. 270n, 1148n, 1262, 1710n Easterly, W. 278, 303, 314n, 326, 331, 565, 567n, 568n, 573, 605, 615, 625, 636, 645n, 653–659, 684n, 781n, 826n, 829, 833, 870n, 923, 953, 971n, 973n, 996, 1005, 1009, 1018n, 1030n, 1032–1034, 1036n, 1045n, 1050n, 1054, 1055, 1505, 1681, 1681n, 1682 Easterly, W., see Alesina, A. 278, 653, 1505, 1505n Easterly, W., see Bruno, M. 655, 1044n, 1048n Eaton, J. 56, 59n, 79n, 493, 647, 711, 820, 822, 835, 837, 841, 842, 854n, 938, 1470n, 1548, 1550 Echevarria, C. 1263 Eckhaus, R.S., see Sue Wing, I. 1795 Eckstein, Z. 227n, 1281, 1281n, 1282, 1282n, 1283, 1283n, 1284, 1285 Eckstein, Z., see Botticini, M. 1155 Eckstein, Z., see Eaton, J. 1548, 1550 Economics and Statistics Administration, 746n, 778n, 807n Edgerton, R.B. 402 Edison, H. 907n Edlund, L. 231
Author Index Edmonds, E. 512 Edwards, S. 653, 663, 1514, 1514n Eeckhout, J. 938 Eggimann, G. 410 Eichengreen, B. 652 Eicher, T. 1721n Eicker, F. 643 Eisenberg, M.J. 1249 Eisenstein, E. 1122n Ekelund Jr., R.B. 1162 Elliott, P. 1149 Ellison, G. 516n, 517, 1556, 1557, 1577 Elston, M., see Powell, C. 498 Endler, J.A. 265n Engerman, C. 1118n Engerman, S.L. 212n, 278, 302, 443, 449, 923, 1005, 1054, 1611, 1617 Engerman, S.L., see Fogel, R.W. 1265 Ensminger, J. 1656 EPA, 1753n Epifani, P. 142n Ergungor, O.E. 919 Erlich, I. 231n Ertman, T. 452 Ervin, L., see Long, J. 643 Esfahani, H. 655, 662 Espinosa, K., see Palloni, A. 1673, 1675, 1680 Esquivel, G., see Caselli, F. 136n, 579, 589, 629, 632, 638, 653, 654, 656, 683, 953, 1048, 1446n Estevadeordal, A. 216, 1532, 1532n Eswaran, M. 508 Ethier, W.J. 113, 116, 1082 Evans, E.J. 459 Evans, G. 149n, 158, 162, 164n, 165 Evans, P. 77, 599, 603 Evenson, R.E. 78n, 852n Fafchamps, M. 480, 513, 1644, 1646, 1646n, 1650, 1651, 1651n, 1653–1655, 1673, 1674, 1686, 1692 Fafchamps, M., see Barr, A. 1657 Fafchamps, M., see Bigsten, A. 1650 Fafchamps, M., see Durlauf, S.N. 402 Fairchilds, C. 1118n Fajnzylber, P., see Loayza, N.V. 978n Faletto, E., see Cardosso, F.H. 428 Fama, E. 872n Fan, J., see Claessens, S. 874 Fan, X., see Thomas, V. 1724n Farber, H.S. 1326
I-9 Farrell, J. 428n Farrington, B. 1134 Fatás, A. 36, 37, 45, 47 Fattouh, B., see Deidda, L. 653 Fay, M. 1560 Faye, M., see Sachs, J.D. 298, 300n, 304n Fazzari, S. 511n Feder, B. 1198 Feder, G. 659 Federal Reserve Bank of Dallas, 1258n Federal Reserve Bank of St. Louis, 1213n Feenstra, R.C. 711n, 1425n Fei, J.C.H. 1713 Fei, J.C.H., see Rannis, G. 1577 Feinstein, C.H. 197, 255 Feinstein, C.H., see Matthews, R.C. 195, 197 Feinstein, J.S., see Casella, A. 1510n Feld, L. 656 Feldman, M.W., see Cavalli-Sforza, L.L. 266n Fellman, J. 1603n Fernald, J., see Basu, S. 1296 Fernandez, C. 587, 587n, 612, 613, 613n, 614, 1681 Fernandez, R. 259n, 262n, 1268, 1603n, 1606, 1678 Fernandez-Villaverde, J. 227, 228, 236n, 255, 255n, 258, 261, 278, 280, 1238n, 1262 Ferreira, F. 1737n Ferreira, F., see Bourguignon, F. 1730n, 1737n Fershtman, C. 1655 Feynman, R.P. 1009 Feyrer, J. 256n, 332n, 606, 707n Fiaschi, D. 256n, 598 Field, A. 212n Fields, G.S. 1713n, 1732, 1734n Figini, P. 1609 Findlay, R. 217, 281n, 1425n, 1510n Finegan, T.A., see Margo, R.A. 1249 Fink, G. 897 Fischer, S. 654–656, 825n, 1033 Fischer, S., see Dornbusch, R. 143, 1463n, 1468 Fisher, A. 719n Fisher, I. 749 Fisher, O’N. 1425n Fisman, R.J. 889, 904n, 912n Fitschen, A., see Klitgaard, R. 1036 Flam, H. 1425n Flamm, K. 751, 807n Flannery, M. 878n Flatters, F. 1567
I-10 Flinn, C. 1354 Flora, P. 205–207, 209, 211, 215, 226, 255 Flug, K. 1301 Fogel, R.W. 1265 Fogli, A., see Fernandez, R. 262n, 1268 Foray, D., see Cowan, R. 1122n Forbes, K.J. 541, 541n, 542, 653, 655, 1605n Forslid, R., see Baldwin, R. 1425n, 1486n Forster, B.A. 1772, 1777 Fortin, N.M. 1597n Fortin, N.M., see DiNardo, J. 1326 Foster, A.D. 239n, 482, 495–497, 500n, 516, 938 Fournier, M., see Bourguignon, F. 1737n Fox, R. 1138, 1163n Francois, J.F. 1425n Francois, P. 157n Frank, A.G. 428, 1170n Frank, K. 1678 Frankel, J.A. 99, 639, 640, 659, 663, 975, 1032, 1096, 1422n, 1515, 1525 Frankel, M. xin, 1103, 1103n, 1718 Frankenberg, E., see Thomas, D. 490 Franses, P., see Hobijn, B. 599, 601 Frantz, B., see Estevadeordal, A. 216, 1532, 1532n Fraumeni, B.M. 760n Fraumeni, B.M., see Jorgenson, D.W. 760n, 763, 774, 774n, 782n, 783n, 1291 Freeman, R.B. 488n, 496, 1007, 1326n, 1328, 1597n, 1610, 1624 Freeman, R.B., see Borjas, G.J. 142n Freeman, S. 354n Frenkel, J. 1425n Freshwater, D., see Rupasingha, A. 658 Freudenberger, H. 433 Frey, B. 1741n Friedman, D. 1510n Friedman, E. 999 Friedman, J. 620n Friedman, J., see Breiman, L. 619n Friedman, J., see Thomas, D. 490 Friedman, M. 593, 599, 1354 Fudenberg, D. 1659 Fudenberg, D., see Ellison, G. 516n Fujita, J. 1578, 1579 Fujita, M.P. 1021, 1486n, 1565, 1576, 1579 Fujita, M.P., see Abdel-Rahman, H. 1564, 1574 Fukao, K. 1485n Fukuyama, F. 1643
Author Index Fullerton, D., see King, M.A. 847n Funke, M. 145n Furfine, C.H. 885 Furstenberg, F. 1660, 1676–1678 Futia, C.A. 311n, 376 Gabaix, X. 1554, 1576 Gabaix, X., see Chaney, T. 1342 Gale, D. 874, 1425n Gale, D., see Allen, F. 876, 881, 882, 884, 886, 918, 918n, 920 Gale, L. 1766n, 1769 Galetovic, A. 871 Gallet, C.A., see List, J.A. 1715 Gallup, J.L. 655, 662, 1560 Galor, O. 92n, 176n, 177, 193n, 194n, 212n, 224n, 228–231, 231n, 232, 235n, 236, 236n, 237, 239n, 255n, 256n, 257–259, 259n, 260–264, 266, 267n, 270n, 272n, 274, 275, 278–282, 351, 354n, 398, 441, 537, 583n, 872, 878, 887, 1098n, 1099, 1118n, 1146, 1261, 1264, 1266, 1267, 1302, 1378, 1425n, 1571, 1604n, 1605n, 1613n, 1720, 1721n Gambera, M., see Cetorelli, N. 913 Gancia, G. 130, 143, 144, 1425n Gancia, G., see Epifani, P. 142n Garby, L., see Li, R. 490 Garcia-Peñalosa, C., see Aghion, P. 1715n, 1721, 1721n Garcia-Peñalosa, C., see Checchi, D. 356 Garicano, L. 1321 Garner, P. 1513n Garrard, J. 460 Gates, B. 1193n Gatti, R., see Dollar, D. 661 Gauthier, B., see Bigsten, A. 1650 Ge, Y., see Anderson, G. 595 Geertz, C. 1656 Geertz, H., see Geertz, C. 1656 Gelb, A. 711n Gelos, R.G. 507, 508 Gennaioli, N., see Caselli, F. 507, 535, 543 Germidis, D., see Lecaillon, J. 1732 Geroski, P.A. 86, 89, 939n Gerschenkron, A. 79, 99, 433, 882, 937, 971 Gertler, M. 1490n Gertler, M., see Bernanke, B.S. 82n Gertler, P.J. 498, 513 Gertler, P.J., see Banerjee, A.V. 508, 509 Gertler, P.J., see Miguel, E. 1687 Ghatak, M. 537
Author Index Ghatak, M., see Banerjee, A.V. 508, 509 Ghate, P. 479, 480 Giavazzi, F. 636, 636n, 646 Gilbert, C.L. 985n Gilligan, D., see Pargal, S. 1675 Gillispie, C.C. 1132n, 1139 Gintis, H., see Bowles, S. 212n, 259n, 1643, 1650n Glaeser, E.L. 278–280, 425, 426, 1020n, 1553, 1586, 1667, 1689 Glaeser, E.L., see Ades, A.F. 1021, 1422n, 1518, 1560, 1561, 1563, 1564 Glaeser, E.L., see Alesina, A. 1597n, 1609 Glaeser, E.L., see DiPasquale, D. 1686 Glaeser, E.L., see Djankov, S. 1006–1008 Glaeser, E.L., see Ellison, G. 517, 1556, 1557, 1577 Glass, A.J. 1425n Glick, R., see Aizenman, J. 654, 655 Glomm, G. 1603n Glynn, P.W. 311n Gneezy, U., see Fershtman, C. 1655 Goddard, N. 1138n Goetz, S. 590 Goetz, S., see Rupasingha, A. 658 Goldin, C. 198, 212, 214, 233n, 257n, 1205, 1250n, 1266, 1267, 1267n, 1268, 1282, 1309, 1309n, 1310, 1311, 1311n, 1360, 1682, 1684, 1725n Goldin, C., see Katz, L.F. 1205, 1205n Goldsmith, R.W. 781, 867, 881, 889, 890, 918, 918n Goldstein, M. 482, 495, 508 Goldstone, J.A. 997, 1171n Golinski, J. 1138, 1139 Gollin, D. 721, 723n, 731, 833, 949n, 1387 Gollin, D., see Evenson, R.E. 852n Gollop, F.M. 763n Gollop, F.M., see Jorgenson, D.W. 725, 760n, 763, 774, 774n, 782n, 783n, 1291 Goodfriend, M. 222n, 1099 Gordon, R.J. 746n, 751, 752, 783n, 806n, 1185, 1196n, 1292, 1295n, 1298n, 1334n, 1360 Gordon, R.J., see Baily, M.N. 779n Gort, M. 1206 Gort, M., see Bahk, B.H. 1204, 1294, 1295n, 1296 Gorton, G. 878n Gosling, A. 1326
I-11 Gottschalk, P. 1286, 1331, 1336, 1342, 1350, 1356, 1733 Gottschalk, P., see Dooley, M. 1319n Gould, E.D. 231, 1336, 1337, 1359 Gourinchas, P.-O. 687n Graddy, K., see Barret, S. 1798n Gradstein, M. 367n, 1616n, 1741n Gradstein, M., see Justman, M. 1741 Gradus, R., see Smulders, S. 1804, 1805 Graham, B.S. 256, 337, 607n, 623, 723n, 727 Graham, C. 1692n Granato, J. 658 Grandmont, J.-M. 164n Granger, C.W.J. 603 Granovetter, M. 1644, 1646, 1647, 1653, 1692 Grant, B.R. 265n Grant, P.R., see Grant, B.R. 265n Grantham-McGregor, S., see Powell, C. 498 Graves, M.A.R. 452 Greasley, D. 601 Green, A. 197, 207, 207n, 208–211 Green, D.A., see Beaudry, P. 660, 1304 Green, J., see Scotchmer, S. 144, 145 Greenbaum, S.J., see Boot, A.W.A. 882 Greenspan, A. 746n Greenston, P., see Bergsman, J. 1555 Greenwood, J. 150n, 229n, 262n, 541n, 715, 768, 783n, 871, 876, 880, 881, 888, 1098n, 1106, 1185, 1194, 1204, 1207, 1227n, 1230n, 1231n, 1238n, 1245n, 1252n, 1262n, 1264n, 1291, 1292, 1294–1296, 1301, 1302, 1312, 1312n, 1338, 1425n Gregory, P.R. 433 Greif, A. 298n, 401, 1117, 1171, 1172, 1653 Grier, K.B. 33, 581n Griffith, R. 78n Griffith, R., see Acemoglu, D. 102n Griffith, R., see Aghion, P. 89, 90, 102n, 107n, 1504n Griffith, R., see Blundell, R. 76, 86, 89 Griffiths, J. 1128n, 1159n Griliches, Z. 133n, 782, 782n, 806n, 839n, 847, 938, 1069, 1198, 1205, 1298 Griliches, Z., see Jorgenson, D.W. 759, 760n, 782, 783, 783n, 784 Grilli, V. 661 Grimm, B.T. 748, 750, 752 Grimm, B.T., see Parker, R.P. 754 Grootaert, C. 1648, 1673, 1674 Gross, N. 433 Grossman, E., see Kendrick, J.W. 781n
I-12 Grossman, G.M. xiin, 71n, 78n, 114, 121n, 123, 127n, 128n, 129, 145n, 223, 281n, 388, 499, 501, 822, 937, 938, 1069, 1070, 1073, 1087n, 1089, 1090, 1090n, 1092, 1103, 1106, 1414, 1425n, 1504, 1517n, 1576, 1598n, 1619, 1623, 1627n 1757n, 1765, 1766n, 1769, 1770n, 1800n, 1801n, 1810 Grossman, H.I. 259n, 398, 399, 429n Grossman, P.J. 1741n Grossman, S.J. 422, 872, 873, 882, 896 Grove, A.S. 807n Gruber, J., see Gertler, P.J. 513 Guagnini, A., see Fox, R. 1163n Guaresma, J. 654, 661 Guesnerie, R., see Azariadis, C. 164n Gugerty, M.K. 521n, 1687 Guidotti, P., see de Gregorio, J. 654 Guiso, L. 896, 896n, 908, 923, 1678, 1682, 1684 Gullickson, W. 774 Gunderson, M. 1730n Gunning, J.W., see Bigsten, A. 1650 Gunning, J.W., see Collier, P. 567n Gurgand, M., see Bourguignon, F. 1737n Gurkaynak, R.S., see Bernanke, B.S. 731, 731n Gurley, J.G. 867, 875 Gurr, T.R., see Ross, J.I. 450n Gutierrez, H. 417 Gylfason, T. 1030n Gyourko, J., see Glaeser, E.L. 1586 Ha, J. 92n, 96n, 1517 Habakkuk, H.J. 133n Haber, S.H. 442, 443, 885, 887, 888, 908, 909, 923 Haber, S.H., see Maurer, N. 885 Habicht, J.-P., see Thomas, D. 490 Haddad, L., see Maluccio, J. 1675 Haefke, C., see Den Haan, W. 1346 Hagan, J. 1676, 1678, 1679 Haggard, S. 994n, 998n Hahn, J. 633, 634n Haider, S. 1331n Haiss, P., see Fink, G. 897 Hall, A.R. 618, 1131, 1132n Hall, B.H. 839n, 847 Hall, R.E. 136n, 277, 277n, 314, 314n, 402, 605, 606, 643, 652, 654, 658, 662, 684, 685, 688, 751n, 782n, 806n, 829, 833, 938, 1005, 1018n, 1032, 1097, 1294, 1296, 1325, 1390n, 1445, 1466n
Author Index Hall, S. 584 Hallinan, M. 1680n Hamermesh, D.S. 131n, 732, 1297n, 1307n Hamilton, B., see Mills, E. 1553 Hamilton, G., see Clark, G. 182n, 228, 235, 266, 271 Hammour, M.L., see Caballero, R.J. 1204, 1350 Hamoudi, A. 655 Hansen, B. 620 Hansen, G.D. 176n, 236n, 262–264, 264n, 1099, 1238n, 1264, 1374–1376, 1380, 1383, 1386, 1389, 1396 Hansen, H. 1350 Hansen, J., see Rosenstone, S. 1608 Hansen, L.P., see Browning, M. 235n, 1739n Hanson, G., see Harrison, A. 1515n Hanushek, E.A. 239n, 656, 700, 702n, 704, 705n, 938, 1018n Harbaugh, W. 1767n Harberger, A.C. 560, 616, 768, 774, 969 Harchaoui, T.M., see Baldwin, J.R. 786n Hardin, R. 450n Harley, C.K., see Crafts, N.F.R. 255 Harper, M.J., see Dean, E.R. 784n Harper, M.J., see Gullickson, W. 774 Harrigan, J. 718 Harris, C., see Aghion, P. 86, 89, 398 Harris, D. 662 Harris, J.R. 725n, 1127n, 1132n, 1156n, 1158, 1577 Harrison, A. 655, 657, 659, 1515n Harrison, G.W. 985n Harrison, P. 875 Harrod, R. 779 Hart, O.D. 429, 512 Hart, O.D., see Grossman, S.J. 422, 873, 882 Hartwick, J.M. 1788 Harvey, C.R., see Bekaert, G. 653, 661, 907 Hasan, I., see Berger, A.N. 897 Hassler, J. 239n, 1597n, 1612n, 1617n Hatta, T. 985n Hausman, J., see Hahn, J. 634n Hausmann, R. 646, 990, 992, 999, 1000, 1003 Hautvast, J.G., see Li, R. 490 Hayashi, F. 768 Hayek, F.A. 1319, 1645 Hazan, M. 230, 230n, 231n, 236n, 259–262, 1266 Head, K. 1547 Heal, G., see Dasgupta, P. 1752n
Author Index Healy, R., see Bergsman, J. 1555 Heathcote, J. 1282n, 1331n, 1355–1357, 1425n Hecht, J. 752n Heckman, J.J. 710n, 1318, 1319, 1354–1357, 1360, 1739n Heckman, J.J., see Browning, M. 235n, 1739n Heckman, J.J., see LaLonde, R.J. 1360 Heilbron, J.L. 1151n Helleiner, G.K. 986n Helliwell, J. 658, 1504n, 1664, 1682, 1689 Hellmann, T. 981, 982 Hellmann, T., see Da Rin, M. 340, 343 Hellwig, M. 883 Hellwig, M., see Gale, D. 874 Helpman, E. 124, 128–130, 157, 807n, 937, 1107, 1200, 1207, 1337, 1425n, 1446n, 1483n, 1555, 1578 Helpman, E., see Coe, D.T. 79n, 842, 852n, 938 Helpman, E., see Flam, H. 1425n Helpman, E., see Grossman, G.M. 71n, 78n, 114, 121n, 127n, 128n, 129, 145n, 223, 281n, 388, 499, 501, 822, 937, 938, 1069, 1070, 1073, 1087n, 1089, 1090, 1090n, 1092, 1103, 1106, 1414, 1425n, 1504, 1517n, 1576, 1800n, 1801n Helsley, R. 1564, 1568, 1573, 1574, 1586 Henderson, D. 605, 605n, 606, 607 Henderson, J.V. 661, 1556, 1560, 1562–1564, 1566, 1568, 1570, 1571, 1573, 1574, 1579, 1582, 1585 Henderson, J.V., see Au, C.C. 1585 Henderson, J.V., see Black, D. 1548, 1550, 1553–1555, 1558, 1564, 1569, 1571, 1577 Henderson, J.V., see Davis, J. 1560, 1561, 1564 Henderson, J.V., see Flatters, F. 1567 Henderson, S.G., see Glynn, P.W. 311n Hendricks, L. 684, 833, 1380n, 1390n Hendry, D. 612, 640 Henrekson, M., see Lybeck, J.A. 1741n Henry, P. 636, 646, 907n Herbst, J.I. 432, 1506n Hercowitz, Z. 768 Hercowitz, Z., see Flug, K. 1301 Hercowitz, Z., see Greenwood, J. 715, 768, 1106, 1291, 1292, 1294, 1295 Hernandez, D.J. 233 Hernandez, P., see Sandefur, G. 1677, 1679 Herrendorf, B. 1395n
I-13 Heston, A. xii, 303n, 491, 562, 685, 804n, 826, 828, 830, 831, 834, 1412, 1519 Heston, A., see Kravis, I.B. 803 Heston, A., see Summers, R. 582, 685, 720, 804, 1037, 1403–1406 Hibbs, D.A. 277n Hickman, W.B. 1214 Hicks, J.R. 133n, 398, 440, 877 Hicks, L., see Caplow, T. 1251 Higgins, M. 660 Higgins, R.C. 915 Higuchi, Y., see Mincer, J. 1317, 1318 Hilaire-Pérez, L. 1133n, 1142n, 1165n Hill, C. 402, 454, 456 Hill, R. 721 Hills, R.L. 1139n Hilton, H. 1766n, 1767n, 1768 Hilton, R. 427 Hirschman, A.O. 971, 998 Hirshleifer, D., see Bikhchandani, S. 516n Hirshleifer, J. 398 Hitt, L.M., see Bresnahan, T.F. 1322 Hitt, L.M., see Brynjolfsson, E. 806n, 1322n Ho, M.S. 1294n Ho, M.S., see Jorgenson, D.W. 779, 782n Hobijn, B. 599, 601, 1207, 1215n Hobjin, B. 1293 Hobsbawm, E. 1532, 1532n Hochman, O. 1564, 1571 Hodges, J., see Draper, D. 582 Hoeffler, A. 629 Hoeffler, A., see Bond, S. 632 Hoff, K. 299, 303, 330, 364, 367n, 368, 369, 480n, 999, 1001 Hoff, K., see Bowles, S. 303 Hofferth, S. 1687 Hoffmaister, A.W., see Coe, D.T. 79n, 852n Hoffman, P.T. 1387 Holmes, T.J. 1300n, 1310n, 1414, 1574 Holmstrom, B. 872, 878, 895 Holtz-Eakin, D. 632, 1818n Homer, S. 1213n, 1220n Honkapohja, S., see Evans, G. 149n, 158, 162, 164n, 165 Hoover, K. 611, 612, 643 Hoover’s, Inc., 1210n Hopenhayn, H.A. 1208 Hopkins, D.R. 1169n Horn, J. 1128, 1168
I-14 Hornstein, A. 1204, 1295, 1295n, 1296, 1297n, 1307, 1309n, 1332, 1333, 1346, 1347, 1347n, 1349, 1350, 1393n Horowitz, D.L. 432 Horrell, S. 230n, 260n Hoshi, T. 884n, 918n Hotelling, H. 1790n Howitt, P.W. 77, 81, 93n, 277n, 500n, 647, 820, 822, 835, 836n, 837, 839, 841, 843, 939, 1094, 1453n, 1517, 1681 Howitt, P.W., see Aghion, P. xiin, 16n, 28, 69, 70, 71n, 82, 86, 89, 90, 102n, 106n, 114, 144, 157, 157n, 223, 256n, 388, 398, 499, 501, 510n, 529, 820, 822, 872, 899, 1018, 1020n, 1069, 1070, 1073, 1087n, 1089, 1090, 1090n, 1092, 1103, 1106, 1116n, 1296, 1302, 1313, 1314n, 1315, 1332, 1333, 1347, 1504n, 1517n, 1721, 1770, 1770n, 1772, 1788, 1800n, 1801n, 1810n Howitt, P.W., see Ha, J. 92n, 96n, 1517 Hoxby, C.M. 1324n Hoxby, C.M., see Aghion, P. 102n Hoxby, C.M., see Alesina, A. 1505n Hristoforova, S., see Fink, G. 897 Hsieh, C.-T. 493, 494, 687n, 728n, 1019n Hu, D., see Goetz, S. 590 Huang, F. 1659 Hubbard, G., see Fazzari, S. 511n Hubbard, T.N., see Baker, G.P. 1322, 1325 Hudson, D. 1142n Huggett, M. 1295n, 1356 Hughes, K., see Rodgers, J.L. 269n Hughes, M., see Furstenberg, F. 1660, 1676–1678 Hultberg, P. 663 Hulten, C.R. 655, 716n, 759n, 782n, 783n, 1292 Human Mortality Database, 204 Hummels, D. 835 Humphries, J. 1128 Humphries, J., see Horrell, S. 230n, 260n Hung, V.T.Y., see Blackburn, K. 871 Huntington, S.P. 1741n Huq, M., see Pargal, S. 1675 Hurt, J. 209 Hussein, K., see Demetriades, P.O. 905, 906 Huybens, E. 886, 923 Hwang, J., see Blattman, C. 656 Hyslop, D. 1358 Ichimura, H., see Attanasio, O. 1356 Ichino, A., see Bertola, G. 1336n, 1342
Author Index Imbs, J. 157, 1000n Impullitti, G., see Cozzi, G. 1308 Inglehart, R., see Granato, J. 658 Ingram, B. 1302 Inkster, I. 1144 Inoue, A., see Diebold, F. 603 International Monetary Fund, 1017 Ioannides, Y.M. 1548, 1553, 1554 Ioannides, Y.M., see Dobkins, L.H. 1548, 1558 Ioannides, Y.M., see Henderson, J.V. 1564, 1573 Irwin, D.A. 807n, 1295n Isaksson, A., see Bigsten, A. 481, 1650 Isham, J. 1673, 1674 Ishii, T., see Fujita, M.P. 1576 Islam, N. 582n, 586n, 628, 629, 635, 635n, 683, 786, 804n, 805, 806, 938 Israel, J.I. 454 Iversen, T. 1326 Iyer, L., see Do, T. 508 Iyigun, M.F. 259n Jacklin, C. 878 Jackson, M. 1660 Jacob, M.C. 1137n Jacoby, H.G. 498, 878 Jaffe, A.B. 1763n, 1798 Jaffe, A.B., see Newell, R.G. 1796 Jakubson, G.H., see Fields, G.S. 1732 Jalan, J. 339, 1035 Jalan, J., see Ravallion, M. 1035 James, H. 1164n Jamison, D., see Bhargava, A. 655 Jayaratne, J. 907, 907n Jean, Y., see Carson, R.T. 1813 Jeanne, O., see Gourinchas, P.-O. 687n Jensen, M. 872n, 873, 874, 882 Jensen, M., see Fama, E. 872n Jensen, R. 1425n Jeong, H. 543 Jimenez, E. 1658 John, A., see Cooper, R.W. 145 Johnson, D.G. 1355, 1385 Johnson, G.E. 703n Johnson, M., see Zhang, X. 1706n Johnson, P.A. 332n, 597, 598, 606, 620n Johnson, P.A., see Durlauf, S.N. 77n, 256, 331, 579n, 589, 607, 619, 619n, 622, 942, 1681 Johnson, P.A., see Temple, J.W. 658, 659 Johnson, S. 508, 914, 999, 1653
Author Index Johnson, S., see Acemoglu, D. 193n, 277, 278, 302, 393, 397n, 402, 410, 413, 414, 417, 417n, 419, 453, 464, 506, 639n, 651, 654, 656–658, 662, 923, 1005, 1033, 1054, 1055 Johnson, S., see Aslund, A. 989, 999 Johnson, S., see Friedman, E. 999 Johnson, W.R., see Neal, D.A. 704, 705n Jolly, R., see Chenery, H.B. 1714n, 1724n Jones, C.I. xin, 92, 93, 98, 117n, 176n, 219, 236n, 256, 261, 563, 626, 822, 838, 841, 843, 857, 1079, 1079n, 1087, 1089, 1091–1093, 1094n, 1095, 1098n, 1099, 1100, 1100n, 1102, 1103n, 1105n, 1120, 1147, 1264, 1305n, 1504n, 1517, 1517n, 1781n Jones, C.I., see Bernard, A. 604n Jones, C.I., see Hall, R.E. 136n, 277, 277n, 314, 314n, 402, 605, 606, 643, 652, 654, 658, 662, 684, 685, 688, 829, 833, 938, 1005, 1018n, 1032, 1097, 1390n, 1445, 1466n Jones, C.P., see Wilson, J.W. 1207n Jones, E.L. 277, 397, 1171 Jones, L.E. 16n, 19, 20, 23, 23n, 39, 41n, 60, 61, 61n, 586n, 820, 823, 825, 1103n, 1266, 1267, 1772, 1798n Jones, L.E., see Boldrin, M. 234n, 1265 Jones, R., see McEvedy, C. 410 Jorgenson, D.W. 682n, 719n, 725, 757n, 759, 759n, 760, 760n, 763, 763n, 769, 771, 774, 774n, 779, 782, 782n, 783, 783n, 784, 785n, 786n, 787, 788n, 805, 870, 1290n, 1291, 1292n, 1294, 1346, 1352n, 1786n Jorgenson, D.W., see Christensen, L.R. 684n, 781n, 787, 805 Jorgenson, D.W., see Conrad, K. 718n Jorgenson, D.W., see Dougherty, C. 787, 805 Jorgenson, D.W., see Ho, M.S. 1294n Jovanovic, B. 500n, 525, 715, 1185, 1189, 1204, 1206, 1207, 1210, 1295n, 1296, 1300, 1333, 1335n, 1344 Jovanovic, B., see Atje, R. 894 Jovanovic, B., see Eeckhout, J. 938 Jovanovic, B., see Greenwood, J. 150n, 541n, 783n, 871, 876, 888, 1185, 1207, 1338 Jovanovic, B., see Hobijn, B. 1207, 1215n Judd, K.L. 113, 122n, 145n, 1070n Judson, R.A. 634 Judson, R.A., see Schmalensee, R. 1818n Juhn, C. 1285, 1286, 1288n Jung, W.S. 905 Junius, K. 1561
I-15 Justman, M. 1741 Justman, M., see Gradstein, M. 1616n, 1741n Kaganovich, M. 20 Kahin, B., see Brynjolfsson, E. 807n Kahn, C., see Calomiris, C. 878n Kahn, L.M., see Bertola, G. 1312, 1340 Kahn, L.M., see Blau, F.D. 1266, 1267, 1267n, 1731 Kahn, M., see Costa, D. 1678, 1680, 1685, 1686 Kalaitzidakis, P. 610 Kalemli-Ozcan, S. 157, 227n Kambourov, G. 1335n, 1336, 1337 Kanbur, R. 513 Kanbur, R., see Anand, S. 1713n, 1732 Kandori, M. 1653, 1658 Kanemoto, Y. 1564, 1573 Kaplan, H.S., see Robson, A.J. 276n Karlan, D.S. 521n Karlan, D.S., see Ashraf, N. 522 Karlan, D.S., see Bertrand, M. 522 Karyadi, D., see Basta, S.S. 490 Kaserer, C., see Wenger, E. 884, 885, 918n Kashyap, A. 878n Kashyap, A., see Hoshi, T. 884n, 918n Katsoulacos, Y., see Beath, J. 349 Katz, L.F. 777n, 807n, 1205, 1205n, 1281, 1281n, 1282, 1285, 1297–1299, 1715n, 1720n, 1724, 1727n Katz, L.F., see Autor, D.H. 1283n, 1597n Katz, L.F., see Borjas, G.J. 142n Katz, L.F., see Goldin, C. 198, 212, 257n, 1205, 1682, 1684, 1725n Katz, L.M., see Cutler, D.M. 1355 Katz, L.M., see Goldin, C. 1282, 1309, 1309n, 1310, 1311, 1311n, 1360 Kauffman, S.A. 1147 Kaufman, R.F. 1797n Kaufman, R.F., see Haggard, S. 998n Kaufmann, D. 649, 974n Kaufmann, D., see Friedman, E. 999 Kawachi, I., see Lochner, K. 1683 Keefer, P. 658, 662 Keefer, P., see Knack, S. 402, 403, 506, 649, 658, 1644, 1660, 1682 Keele, K.D. 1169 Keeler, E. 1774n, 1786, 1810n Keeler, M. 455 Kehoe, P.J., see Atkeson, A.A. 1098n, 1204, 1296, 1325, 1360, 1452n
I-16 Kehoe, P.J., see Backus, D.K. 1096, 1425n, 1517 Kehoe, P.J., see Chari, V.V. 1019n Kehoe, T.J. 350 Kehoe, T.J., see Backus, D.K. 1096, 1517 Keirzkowsky, H., see Findlay, R. 281n Keller, W. 79n, 852, 857, 949n Kelley, A. 657 Kelley, E.M. 1210n, 1212 Kelly, A.C. 1578 Kelly, F.C. 1153n Kelly, M. 586n, 1359 Kelly, T. 654 Kendrick, J.W. 781, 781n, 784, 1218n Kennedy, C. 131n Kenny, C. 1032 Kettlewell, H.B.D. 265n Keyser, B.W. 1139 Keyssar, A. 448 Khan, B.Z. 1165, 1166 Khan, M., see Luintel, K.B. 906n Kierzkowski, H., see Findlay, R. 1425n Kihlstrom, R. 513 Kiiski, S., see Cornia, G.A. 1733 Kiley, M.T. 806n, 1305, 1598n, 1620n Killick, T. 447 Kim, H.S. 1565 Kim, M., see Grossman, H.I. 259n, 398, 399 Kim, S. 1555, 1556 Kim, S., see Nadiri, M.I. 938 Kimko, D.D., see Hanushek, E.A. 656, 702n, 704, 705n, 938, 1018n King, I. 1332n King, M.A. 847n King, R.G. 314n, 316n, 653, 661, 684, 781n, 870n, 871, 876, 880, 890–893, 895n, 898 Kingdon, G.G., see Dreze, J. 498 Kirby, P. 1155 Kirkwood, T.B.L. 275n Kiviet, J. 634 Kiviet, J., see Bun, M. 634 Kiyotaki, N. 350 Klapper, L.F., see Berger, A.N. 897 Klasen, S. 661 Klein, M. 907n Klenow, P.J. 135, 136n, 499, 604–606, 627, 646n, 647, 648, 684, 688, 689, 698, 706, 833, 834, 938, 948, 1018n, 1032, 1390n, 1445, 1466n Klenow, P.J., see Bils, M. 495, 502, 653, 687n, 698, 706, 833, 938, 1104n
Author Index Klenow, P.J., see Heckman, J.J. 710n Klenow, P.J., see Hsieh, C.-T. 493, 687n, 1019n Klenow, P.J., see Hummels, D. 835 Klenow, P.J., see Irwin, D.A. 807n, 1295n Klepper, S. 591, 642 Kline, P., see Charles, K. 1686 Klitgaard, R. 1036 Knack, S. 402, 403, 506, 649, 656, 658, 1644, 1660, 1682 Knack, S., see Keefer, P. 658, 662 Knack, S., see Zak, P. 658, 1658, 1683 Knight, J.B. 1713n Knight, J.B., see Gelb, A. 711n Knight, M. 579, 683, 1446n Knight, M., see Loayza, N.V. 1425n Knowles, K. 1158 Knowles, S. 653, 655 Kocherlakota, N.R. 626, 1355, 1414 Kogel, T. 236n, 262 Koh, W., see Baltagi, B. 644 Kohler, H. 269n Kohler, H., see Rodgers, J.L. 269n Kolko, J. 1556, 1574 Komlos, J. 223n Kongsamut, P. 1263 König, W. 1165 Konus, A. 750 Koopmans, T.C. xi, 16, 17, 298, 388 Koren, M. 718n Kormendi, R.L. 32, 45, 581n, 654–657, 659 Kortum, S.S. 820, 823, 1089, 1105, 1199 Kortum, S.S., see Eaton, J. 79n, 493, 647, 711, 820, 822, 835, 837, 841, 842, 854n, 938, 1470n Kosmin, B.A. 1035 Kotwal, A., see Eswaran, M. 508 Kourtellos, A. 618, 621 Kourtellos, A., see Durlauf, S.N. 617, 618, 1681 Kraay, A. 1490n Kraay, A., see Dollar, D. 652, 658, 659, 1422n, 1734, 1736 Kraay, A., see Driscoll, J. 644 Kraay, A., see Kaufmann, D. 649 Kramarz, F., see Abowd, J.M. 1331n Kranakis, E. 1141 Kranton, R. 1646n Kranton, R., see Akerlof, G. 1655 Krasker, W. 649n
Author Index Kraus, F., see Flora, P. 205–207, 209, 211, 215, 226, 255 Kravis, I.B. 803 Krebs, T. 61, 62n, 876n Kreiner, C.T., see Dalgaard, C.-J. 165, 224n Kremer, M. 221n, 223, 224n, 239n, 272n, 340, 346, 497, 498, 596, 597, 827, 857, 1021, 1088, 1095, 1098–1100, 1147, 1324, 1598n, 1619, 1621n Kremer, M., see Duflo, E. 482, 495, 516, 521, 522 Kremer, M., see Easterly, W. 326, 565, 568n, 625, 655, 656, 659, 829, 996, 1036n Kremer, M., see Gugerty, M.K. 1687 Kremer, M., see Miguel, E. 490, 498, 517, 521 Kreps, D. 1602n Krishna, A. 1674, 1676 Kroft, K. 34, 35 Krolzig, H.-M., see Hendry, D. 612, 640 Kronick, D.A. 1134n, 1163 Kroszner, R., see Cowen, T. 910n Krueger, A.B. 491, 540n, 653, 661, 705n, 938, 1018n, 1331n, 1603n, 1724n Krueger, A.B., see Autor, D.H. 1283n, 1597n Krueger, A.B., see Card, D. 702n Krueger, A.B., see Farber, H.S. 1326 Krueger, A.B., see Grossman, G.M. 1757n, 1765, 1766n, 1769, 1770n, 1810 Krueger, A.B., see Katz, L.F. 777n Krueger, A.O. 973n Krueger, D. 1302, 1355–1357 Krugman, P.R. 143n, 157, 328n, 398, 974, 1021, 1340, 1425n, 1453n, 1485, 1485n, 1550, 1568, 1578, 1579 Krugman, P.R., see Brezis, E.S. 278n Krugman, P.R., see Brezis, S.L. 1427n, 1453n Krugman, P.R., see Fujita, J. 1578, 1579 Krugman, P.R., see Fujita, M.P. 1021, 1486n Krugman, P.R., see Helpman, E. 124, 1446n, 1483n Kruk, M., see Sachs, J.D. 298, 300n, 304n Krusell, P. 131n, 398, 399, 434, 1196n, 1205, 1298, 1304, 1414, 1625n Krusell, P., see Greenwood, J. 715, 768, 1106, 1291, 1292, 1294, 1295 Krusell, P., see Hornstein, A. 1204, 1295, 1295n, 1296, 1297n, 1307, 1332, 1333, 1346, 1347, 1347n, 1349, 1350 Krussel, P. 515, 544n Kubitschek, W., see Hallinan, M. 1680n Kuczynski, P.-P. 974n
I-17 Kuersteiner, G., see Hahn, J. 634n Kugler, M., see Neusser, K. 905 Kuhn, T.S. 175n Kumar, K.B. 917n Kumar, K.B., see Krueger, D. 1302 Kuncoro, A., see Henderson, J.V. 1562, 1563 Kupperman, K.O. 420 Kurian, G.T. 212 Kurlat, S., see Alesina, A. 278, 653, 1505 Kuuluvainen, J., see Tahvonen, O. 1797 Kuznets, S. 217, 434, 719n, 768, 779, 780, 780n, 781, 781n, 782, 782n, 784, 786, 791, 792, 803, 804, 806, 1069, 1201n, 1373, 1704, 1706n, 1712 Kuzynski, R.R. 201 Kwan, F.Y.K. 123 Kydland, F., see Backus, D. 1425n Kyle, A.S. 872 Kyn, O., see Papanek, G. 1733n La Ferrara, E., see Alesina, A. 662, 1505n, 1686, 1688 La Porta, R. 84n, 399, 425, 426, 506, 658, 874, 885, 887, 892, 897, 922, 923, 1033, 1505, 1683 La Porta, R., see Djankov, S. 99, 425, 426, 506, 509n, 1006–1008, 1395 La Porta, R., see Glaeser, E.L. 278–280 Lachman, S.P., see Kosmin, B.A. 1035 Lack, D. 231, 269, 269n Ladron de Guevara, A. 26 Laeven, L. 885 Laeven, L., see Caprio Jr., G. 874, 885, 923n Laeven, L., see Claessens, S. 653, 661, 913, 914 Laeven, L., see Demirgüç-Kunt, A. 907n Laffont, J.-J. 509n Laffont, J.-J., see Kihlstrom, R. 513 Lagerlof, N.-P. 234, 236n, 238, 255, 259, 261, 262, 278, 280 Lagerlof, N.-P., see Edlund, L. 231 Lai, E.L.-C. 124, 130, 1425n Lai, E.L.-C., see Grossman, G.M. 123 Lai, E.L.-C., see Kwan, F.Y.K. 123 Laibson, D. 520 Laibson, D., see Glaeser, E.L. 1020n, 1667, 1689 Laitner, J. 1207, 1263 LaLonde, R.J. 1360 Lambsdorff, J.G. 364 Lamo, A. 598, 598n
I-18 Lamo, A., see Andres, J. 598, 598n Lamoreaux, N. 879, 888 Landefeld, J.S. 750n Landes, D. 561 Landes, D.S. 207, 277, 402, 419, 1171 Landes, W. 1249 Lang, L., see Claessens, S. 874 Lang, S. 458, 459 Lapham, B.J., see Devereux, M.B. 124, 126, 126n, 1425n Lasota, A. 377 Lau, L.J. 759n, 980 Lau, L.J., see Bhargava, A. 655 Laumann, E., see Sandefur, R. 1690n Laurini, M., see Andrade, E. 598 Lavezzi, A.M., see Fiaschi, D. 256n, 598 Law, S., see Demetriades, P.O. 653, 661, 662 Lawrence, D.A., see Diewert, W.E. 760n Layard, R., see Nickell, S. 1340, 1341, 1353 Lazear, E.P., see Freeman, R.B. 1328 Le Van, C., see Dimaria, C.H. 350 Leamer, E. 609, 610, 649 Leamer, E., see Klepper, S. 591, 642 Lebergott, S. 1228, 1229, 1246, 1249n, 1258n, 1259n Leblang, D., see Eichengreen, B. 652 Leblang, D., see Granato, J. 658 Leblebicioglu, A., see Bond, S. 627, 660 Lecaillon, J. 1732 Lederman, D. 833, 834, 839, 846, 849, 854 Lee, D. 1299n, 1597n Lee, F., see Coulombe, S. 656 Lee, J.-W. 701, 702, 703n, 704 Lee, J.-W., see Barro, R.J. 105, 195, 216, 283n, 491n, 589, 638, 653, 655–659, 685, 694, 695, 826, 827, 829–831, 834, 846, 948, 1037, 1519 Lee, J.Y., see Henderson, J.V. 1556, 1563 Lee, K.S. 586n, 629, 634, 635, 1562 Lee, R.D. 182n, 272n, 1098–1100, 1710n Lee, S.J. 459, 1673, 1675 Lee, T.C. 1555, 1556, 1561 Lee, T.C., see Henderson, J.V. 1556, 1563 Lee, W. 1609n Lefort, F., see Caselli, F. 136n, 579, 589, 629, 632, 638, 653, 654, 656, 683, 953, 1048, 1446n Legovini, A., see Bouillon, C.P. 1035 Legros, P. 1598n, 1623 Leland, H. 39 Lemieux, T., see Card, D. 1285n Lemieux, T., see DiNardo, J. 1326
Author Index Lemieux, T., see Fortin, N.M. 1597n Leonard, H., see Leamer, E. 609 Leonard, J.S., see Dickens, W.T. 1326 Leonardi, R., see Putnam, R.D. 402, 658, 1643, 1648 Lerner, J. 1165n, 1166, 1199 Lerner, J., see Kortum, S. 1199 Lester, R.K., see Dertouzos, M. 782n Lettau, M., see Campbell, J. 1342 Levchenko, A. 1428n, 1491n Levhari, D. 39, 44, 113n, 870 Levin, R.C., see Cohen, W.M. 89, 136 Levine, D.K., see Boldrin, M. 28, 30, 62, 1088, 1207 Levine, D.K., see Kehoe, T.J. 350 Levine, D.K., see Miguel, E. 1687 Levine, P., see Chui, M. 1425n Levine, R. 84, 441, 493, 575, 609, 610, 645, 647, 648, 653–657, 659, 661, 877, 886, 893–895, 895n, 896, 896n, 897, 898, 898n, 899–901, 907, 907n, 918n, 919, 1033 Levine, R., see Barth, J.R. 922 Levine, R., see Beck, T. 653, 655, 661, 897, 897n, 900, 902, 904, 904n, 907n, 912, 915, 918–920, 922, 923, 923n Levine, R., see Boyd, J.H. 923 Levine, R., see Caprio Jr., G. 874, 885, 923n Levine, R., see Demirgüç-Kunt, A. 881, 894n, 907n, 918, 920 Levine, R., see Easterly, W. 278, 314n, 326, 331, 567n, 605, 615, 645n, 653–659, 684n, 829, 833, 870n, 923, 1005, 1018n, 1032, 1050n, 1054, 1055, 1505, 1681, 1681n, 1682 Levine, R., see Edison, H. 907n Levine, R., see King, R.G. 653, 661, 684, 870n, 871, 876, 890–893, 895n, 898 Levine, R., see Laeven, L. 885 Levinson, A. 1764 Levinson, A., see Andreoni, J. 1799 Levinson, A., see Harbaugh, W. 1767n Levinson, A., see Hilton, H. 1766n, 1767n, 1768 Levinthal, D.A., see Cohen, W.M. 78n Levy, F. 1281, 1281n Levy, F., see Autor, D.H. 1302n Levy, F., see Murnane, R.J. 704, 704n Levy, P.A. 1326n Levy-Leboyer, M. 188, 214 Lewis, A. 1713 Lewis, H.G., see Becker, G.S. 228 Lewis, W.A. 413n, 719n, 1577
Author Index Ley, E., see Fernandez, C. 587, 587n, 612, 613, 613n, 614, 1681 Li, C.-W. 1095 Li, H. 462, 541, 655, 662 Li, Q. 602 Li, R. 490 Li, S. 1005 Li, W. 26 Li, Z., see Lin, J.Y. 980 Licandro, O., see Boucekkine, R. 259 Lichbach, M.I. 450n Lichtenberg, F.R. 656, 1206 Lichtenberg, F.R., see Bartel, A.P. 938, 1205, 1301 Ligon, E., see Conley, T.G. 645 Lillard, L.A. 1318 Limao, N. 300n, 349n Lin, I.-F., see Case, A. 518n Lin, J.Y. 978n, 980 Lin, N. 1644 Lindahl, M., see Krueger, A.B. 540n, 653, 661, 938, 1018n, 1724n Lindauer, D.L. 973n, 1009 Lindbeck, A. 779n, 1321, 1330, 1597n Lindert, P.H. 214, 459, 460, 1724n, 1736 Lindert, P.H., see Williamson, J.G. 1205 Lindquist, M.J. 1300, 1301 Lipscomb, A.A. 1069 Lipset, S.M. 1035 Lipsey, R.G. 1295 Lipsey, R.G., see Blomstrom, M. 627, 656, 659 Lipton, D. 989 List, J.A. 1715 Liu, M., see Lin, J.Y. 978n Liu, Z. 589, 617, 618 Livi Bacci, M. 185, 1148n Livingston, F. 266n Livshits, I., see Bridgman, B. 1414 Ljungqvist, L. 354n, 1342, 1344, 1346 Lleras-Muney, A., see Dehejia, R. 907, 908 Lloyd-Ellis, H. 537n, 1302, 1598n, 1620n Lloyd-Ellis, H., see Francois, P. 157n Lloyd-Ellis, H., see Kroft, K. 34, 35 Loayza, N.V. 661, 903, 978n, 1425n Loayza, N.V., see Beck, T. 900, 902 Loayza, N.V., see Easterly, W. 953 Loayza, N.V., see Knight, M. 579, 683, 1446n Loayza, N.V., see Kraay, A. 1490n Loayza, N.V., see Levine, R. 84, 661, 897, 898, 898n, 899, 900, 1033 Lochner, K. 1683
I-19 Lochner, L., see Heckman, J.J. 1318, 1319, 1354–1357, 1739n Loewy, M. 602 Loh, W.-Y. 620n Londregan, J.B. 661 Londregan, J.B., see Dixit, A.K. 429n Long, B.T., see Hoxby, C.M. 1324n Long, J. 643, 1155 Long, S.K. 498 Lonsdale, J., see Berman, B.J. 449 Lopez, R. 985n, 1770, 1779, 1786, 1786n, 1810 Lopez-de-Silanes, F., see Djankov, S. 99, 425, 426, 506, 509n, 1006–1008, 1395 Lopez-de-Silanes, F., see Glaeser, E.L. 278–280 Lopez-de-Silanes, F., see La Porta, R. 84n, 399, 425, 426, 506, 658, 874, 885, 887, 892, 897, 922, 923, 1033, 1505, 1683 Lora, E. 976, 977, 978n Loury, G.C. 354, 537, 1603n, 1642, 1677 Love, I. 916 Love, I., see Fisman, R.J. 889, 904n, 912n Lowood, H. 1143, 1143n Lucas, R.E. xin, 101, 176n, 221n, 257n, 314n, 316, 388, 491, 558, 807, 819–821, 825, 843, 849, 867, 1079, 1092, 1098n, 1099, 1103, 1104, 1148, 1504, 1546, 1565, 1800, 1801 Lucas, R.E., see Alvarez, F. 1470n Lucas, R.E., see Stokey, N.L. 311n, 320, 376 Lucas Jr., R.E. 23, 24, 30, 62, 329, 768, 1295n, 1336, 1342, 1359, 1378, 1400 Luckhurst, K.W., see Hudson, D. 1142n Lui, F.T., see Erlich, I. 231n Luintel, K.B. 906n Luintel, K.B., see Arestis, P. 905 Lund, S., see Fafchamps, M. 513, 1673 Lundberg, M. 1733n Lundberg, S. 519 Lundblad, C., see Bekaert, G. 653, 661, 907 Lundborg, P. 1442n Lundgren, A. 1151n Lundström, S. 661 Lustig, N., see Bouillon, C.P. 1035 Lustig, N., see Bourguignon, F. 1730n, 1737n Lybeck, J.A. 1741n Lyons, R.K., see Caballero, R.J. 338n Maasoumi, E. 595 MacArthur, R.H. 231, 269 Macfarlane, A. 1152, 1152n
I-20 MacGee, J., see Bridgman, B. 1414 Machin, S. 1288, 1326 Machin, S., see Berman, E. 1597n Machin, S., see Gosling, A. 1326 MacKinnon, J. 643 MacKinnon, J., see Davidson, R. 643 MacLean, B.K. 975, 1001n MacLeod, C. 1165 MacMillan, R., see Hagan, J. 1678 Madalozzo, R., see Andrade, E. 598 Maddala, G. 618 Maddison, A. 76, 175, 179–182, 187–189, 196, 199, 204, 218, 219, 219n, 228n, 256, 307, 307n, 405, 565, 601, 785n, 786, 803, 803n, 804, 969, 1091, 1098, 1117n, 1171, 1373, 1374, 1377n, 1386, 1388, 1398, 1403, 1405, 1410, 1411 Mafezzoli, M., see Cuñat, A. 1425n, 1463n Magendzo, I., see Edwards, S. 653 Maggi, G., see Grossman, G.M. 1598n, 1619, 1627n Magill, M. 373 Mailath, G., see Cole, H. 544 Majluf, N., see Myers, S.C. 872n Majumdar, M. 350 Maksimovic, V., see Beck, T. 914, 916, 917 Maksimovic, V., see Demirgüç-Kunt, A. 886, 889, 914, 916, 917, 919 Malaney, P., see Bloom, D.E. 655 Maler, K.-G. 779n Maler, K.-G., see Dasgupta, P. 1752n, 1760n Malkiel, B., see Campbell, J.Y. 1342 Mallon, R., see Van Arkadie, B. 976n Mallows, C., see Draper, D. 582 Maloney, W.F., see Lederman, D. 839, 854 Malthus, T.R. 221, 221n, 1261, 1374 Maluccio, J. 1675 Maluccio, J., see Carter, M. 1674 Mamuneas, T. 618 Mamuneas, T., see Kalaitzidakis, P. 610 Mankiw, N.G. xii, 77, 477, 502, 578, 586n, 587n, 598, 604, 605, 618, 629, 637, 642, 647, 657, 684, 689, 804, 805, 821, 1018n, 1022n, 1050, 1097, 1445, 1466n Mankiw, N.G., see Barro, R.J. 647 Manova, K., see Aghion, P. 107n, 879 Manovskii, I., see Kambourov, G. 1335n, 1336, 1337 Mansfield, E. 942n Manski, C. 516n, 645, 1643n, 1667, 1668, 1670, 1689
Author Index Mantoux, P. 1168 Manuelli, R.E. 62, 1296, 1333n, 1338 Manuelli, R.E., see Jones, L.E. 16n, 19, 20, 23, 23n, 39, 41n, 60, 61, 61n, 586n, 820, 823, 825, 1103n, 1266, 1267, 1772, 1798n Marcet, A., see Canova, F. 618, 619 Margo, R.A. 1249 Margo, R.A., see Goldin, C. 1311n Margolis, D.N., see Abowd, J.M. 1331n Marimon, R. 1345 Marin, J.M., see de la Fuente, A. 875 Marion, N., see Aizenman, J. 38 Marshall, A. 400, 1546, 1565 Martin, G., see Macfarlane, A. 1152, 1152n Martin, P. 36, 156, 1490n Martin, P., see Baldwin, R. 1425n, 1486n Martin, R., see Corrado, L. 601n Masanjala, W. 612–615 Masanjala, W., see Papageorgiou, C. 619 Maskin, E.S., see Bessen, J. 145 Maskin, E.S., see Dewatripont, M. 884, 920 Maskin, E.S., see Fudenberg, D. 1659 Maskin, E.S., see Kremer, M. 1324, 1598n, 1619, 1621n Maskus, K.E., see Yang, G. 130, 1425n Massey, D., see Palloni, A. 1673, 1675, 1680 Masters, W.E. 277n, 653, 654, 658 Mastruzzi, M., see Kaufmann, D. 649 Mateos-Planas, X. 715 Matoussi, M.S., see Laffont, J.-J. 509n Matsuyama, K. 120n, 143n, 157, 158, 160n, 281n, 303, 318n, 328n, 329, 335n, 349, 351, 354n, 358, 537, 999, 1000, 1425n, 1470n, 1473n, 1485n, 1487n, 1490n Matsuyama, K., see Ciccone, A. 145, 145n Matthews, R.C. 195, 197 Maurer, N. 443, 885 Maurer, N., see Haber, S.H. 443, 887, 923 Mauro, P. 506, 638, 649, 652, 661 May, J., see Maluccio, J. 1675 Mayer, C., see Carlin, W. 920 Mayer, D., see Howitt, P.W. 500n Mayer, T., see Head, K. 1547 Mayer-Foulkes, D. 77, 77n Mayer-Foulkes, D., see Aghion, P. 82, 106n, 256n, 510n, 529, 899 Mayer-Foulkes, D., see Howitt, P.W. 81, 277n, 939, 1681 McArthur, J.W. 654, 656 McArthur, J.W., see Sachs, J.D. 298, 300n, 304n
Author Index McCall, J.J. 1344 McCallum, J. 1504n, 1536 McCarthy, B., see Hagan, J. 1676, 1679 McCarthy, D. 654 McCleary, R., see Barro, R.J. 402, 651, 658, 662 McClellan III, J.E. 1135n, 1143, 1144 McClelland, C.E. 211 McCloy, S.T. 1128n McConnell, K.E. 1770n McConnell, S. 1313n McCord, G., see Sachs, J.D. 298, 300n, 304n McCubbin, D., see Carson, R.T. 1813 McDaniel, T. 433 McDermott, J. 224n, 1414 McDermott, J., see Goodfriend, M. 222n, 1099 McEvedy, C. 410 McFadden, D., see Diamond, P. 733n McGrattan, E.R. 25, 684n, 804n, 1019n, 1325, 1407n McGrattan, E.R., see Chari, V.V. 1019n McGrattan, E.R., see Jones, L.E. 1266, 1267 McGuckin, R. 1206 McKendrick, N. 1131 McKenzie, D. 481, 661 McKinnon, R.I. 867, 909, 910 McKinsey Global Institute, 500, 501 McLanahan, S., see Case, A. 518n McMillan, J. 1653 McMillan, J., see Johnson, S. 508, 914, 999, 1653 McMillan, M.S., see Masters, W.E. 277n, 653, 654 McNeal, R. 1676, 1677, 1679 McNeil, M. 1134n, 1135n Meadows, D.H. 1752n Meadows, D.L., see Meadows, D.H. 1752n Means, G.C., see Berle, A.A. 873 Meckling, W.R., see Jensen, M. 872n, 873, 874 Medoff, J.L., see Freeman, R.B. 1328 Meghir, C. 1331n Meghir, C., see Aghion, P. 102, 104 Meguire, P.G., see Kormendi, R.L. 32, 45, 581n, 654–657, 659 Meier, A., see Sandefur, G. 1677, 1679 Meier, G.M. 867 Melka, J., see Van Ark, B. 786n Mellinger, A., see Gallup, J.L. 655, 662, 1560 Meltzer, A. 1741 Menaghan, E., see Parcel, T. 1679 Mendez, J.A., see Gale, L. 1766n, 1769
I-21 Mendoza, E. 38, 39, 59, 61 Merriman, J.W. 1532 Merton, R.C. 869, 872, 886, 919 Merton, R.K. 1133n Michelacci, C. 602, 603 Mieszkowski, P., see Flatters, F. 1567 Miguel, E. 490, 498, 517, 521, 660, 1687 Miguel, E., see Kremer, M. 498 Milanovic, B. 978n Milesi-Ferretii, G., see Grilli, V. 661 Milgrom, P. 1310, 1320 Milgrom, P., see Greif, A. 298n Miller, M.H. 867 Miller, M.H., see Grossman, S.J. 896 Miller, R., see Doppelhofer, G. 587, 612–615 Miller, W.B., see Rodgers, J.L. 269n Mills, E. 1553, 1560 Mills, E., see Becker, G.S. 1578 Mills, L., see Carlino, G. 602 Mincer, J. 1266, 1317, 1318 Minier, J. 652 Minkin, A., see Durlauf, S.N. 617, 618, 1681 Minten, B., see Fafchamps, M. 1644, 1650, 1651, 1653, 1673, 1674 Mira, P., see Eckstein, Z. 227n Mirman, L.J. 377 Mirman, L.J., see Amir, R. 350 Mirman, L.J., see Brock, W.A. 16, 308n, 320 Mitch, D. 207, 460, 1155 Mitchell, B. 194 Mitchell, M.F. 1310, 1320 Mitchell, M.F., see Holmes, T.J. 1300n, 1310n Mitra, T., see Majumdar, M. 350 Moav, O. 224n, 231n, 1604n, 1719n Moav, O., see Galor, O. 176n, 194n, 212n, 228, 230–232, 235n, 236, 236n, 239n, 258, 259, 259n, 260, 261, 263, 264, 266, 272n, 274, 275, 278–280, 1146, 1302, 1605n, 1613n Moav, O., see Gould, E.D. 231, 1336, 1337 Möbius, M. 1321 Modigliani, F. 1425n Moen, E.R. 1339 Moersch, M., see Black, S.W. 884 Moffitt, R., see Gottschalk, P. 1286, 1331, 1336, 1342, 1356 Mohring, H. 1566 Mokyr, J. 176n, 183n, 197, 207n, 209, 270n, 277, 304n, 306, 434, 561, 1115, 1119, 1123, 1126, 1127, 1129–1131, 1135, 1147, 1162, 1167, 1167n, 1377n Molana, H. 1425n
I-22 Montesquieu, C.S. 400 Montgomery, J. 1653 Montiel, P., see Easterly, W. 953 Moody’s Investors Service, 1210n Mookherjee, D. 259n, 303, 317, 326, 326n, 354n, 537, 537n, 1604n Moore, G.E. 748n, 750n Moore, J.H, see Hart, O.D. 512 Moore, J.H., see Kiyotaki, N. 350 Moore, P., see Fernandez, R. 1678 Moore Jr., B. 425 Morales, M.F. 871 Morck, R. 872–874, 883, 884, 884n, 887, 918n Morduch, J. 340, 357n, 515 Morelli, M., see Ghatak, M. 537 Morenoff, J., see Sampson, R. 1687, 1688 Morgan, D. 885 Morgan, E.S. 448 Morgan, S. 1677, 1679, 1680n Morris, A.G., see Shan, J.Z. 892 Morris, C.T., see Adelman, I. 1706n Morrisson, C. 213, 214, 1736 Morrisson, C., see Bourguignon, F. 1441, 1735 Morrisson, C., see Lecaillon, J. 1732 Morse, C., see Barnett, H.J. 1751n Mortensen, D.T. 1331, 1332, 1346, 1349 Morus, I.R. 1150n Mosse, W.E. 433 Motley, B. 655 Motohashi, K., see Jorgenson, D.W. 786n Moulin, S., see Kremer, M. 497 Moulton, B.R. 754n Mountford, A. 281n, 1425n Mountford, A., see Galor, O. 176n, 193n, 236n, 258, 263, 264, 280–282 Mueller, D.C. 1741n Mueller, E. 1265 Mukand, S. 1007, 1008, 1008n Mulani, D., see Comin, D. 1342 Mulder, N., see van Ark, B. 786n Mullainathan, S., see Bertrand, M. 522 Muller, D.K. 211 Mulligan, C.B. 25, 1103n Munshi, K. 497, 516, 517n Munshi, K., see Banerjee, A.V. 483, 495, 507, 511n, 517 Murdoch, P. 1139 Murdock, K., see Hellmann, T. 981, 982 Murmann, J.P. 1164 Murnane, R.J. 704, 704n Murnane, R.J., see Autor, D.H. 1302n
Author Index Murnane, R.J., see Levy, F. 1281, 1281n Murphy, K.M. 145, 340–342, 364, 365, 398, 503, 506n, 653, 711n, 996n, 999, 1001, 1021, 1285, 1288n, 1359, 1504 Murphy, K.M., see Becker, G.S. 224n, 235, 1021, 1378 Murphy, K.M., see Jensen, M. 874 Murphy, K.M., see Juhn, C. 1285, 1286 Murphy, K.M., see Katz, L.F. 1297–1299 Murphy, M., see Dyson, T. 190n Murray, C., see Bhargava, A. 655 Murshid, K. 479, 480 Musson, A.E. 1128n, 1132 Myaux, J., see Munshi, K. 517n Myers, M.G. 1425n Myers, S.C. 872n Myrdal, G. 400 Nadiri, M.I. 938 Nadiri, M.I., see Hultberg, P. 663 Nagypal, E., see Eckstein, Z. 1281, 1281n, 1282, 1282n, 1283, 1283n, 1284, 1285 Naim, M. 974n Nakamura, M., see Morck, R. 884, 918n Namunyu, R., see Kremer, M. 497 Nanetti, R.Y., see Putnam, R.D. 402, 658, 1643, 1648 Narayan, D. 1666, 1673, 1675 Narayan, D., see Woolcock, M. 1020n Nardinelli, C. 1249, 1265 Nasution, P., see Henderson, J.V. 1562 Neal, D.A. 704, 705n Neal, L. 456 Neary, J.F. 1579 Neary, J.P., see Anderson, J.E. 985n Neary, P. 142n Neher, A.P. 234n Nelson, J.M., see Huntington, S.P. 1741n Nelson, M. 654 Nelson, R.L. 1206n Nelson, R.R. 78, 98, 102, 239, 270n, 373, 939, 1122n, 1162n, 1205, 1301, 1304 Nerlove, M. 632 Neumann, G., see Ingram, B. 1302 Neumark, D. 1286, 1335 Neusser, K. 905 Newell, R.G. 1796 Newey, W., see Holtz-Eakin, D. 632 Newman, A.F. 513 Newman, A.F., see Banerjee, A.V. 92n, 354n, 398, 399, 441, 513, 515, 537, 539n, 544n, 887, 1604n, 1623, 1720, 1722
Author Index Newman, A.F., see Legros, P. 1598n, 1623 Newton, A.P. 420 Ngai, L.R. 1374n, 1402n Ngyen, S., see McGuckin, R. 1206 Nickell, S. 76, 86, 89, 631, 1340, 1341, 1341n, 1350, 1353 Nishimura, K. 308n, 350 Nishimura, K., see Dechert, W.D. 350 Nishimura, K., see Magill, M. 373 Noh, S.J., see Grossman, H. 429n Nordhaus, W.D. 113, 122, 1065n, 1070, 1751n, 1752n, 1755n, 1782 Norman, V., see Dixit, A.K. 1446n North, D.C. 150, 277, 301, 364, 367, 388, 419, 423, 427, 429n, 443, 453, 457, 463, 1005, 1007, 1054, 1117, 1127, 1649, 1650 Nugent, J., see Campos, N. 627 Nunziata, L., see Nickell, S. 1341n Nurkse, R. 297, 719n Nyarko, Y., see Jovanovic, B. 1204, 1295n Nyarko, Y., see Majumdar, M. 350 Ó Gráda, C., see Allen, R.C. 1138n Oates, W.E. 1505n O’Brien, P. 1161n Obstfeld, M. 876, 1425n, 1490n, 1504n Ocampo, J.A. 974n, 998n, 999 Odedokun, M. 653 Odling-Smee, J.C., see Matthews, R.C. 195, 197 Oduro, A., see Bigsten, A. 1650 OECD Employment Outlook, 1288 Ofek, H. 266 Ogawa, H., see Fujita, M.P. 1565 Ogburn, W.F. 1268 Ogilvie, S. 1691 Ohanian, L.E. 1098n Ohanian, L.E., see Krusell, P. 131n, 1196n, 1205, 1298, 1304 Ohlsson, H., see Agell, J. 662 Ohm Kyvik, K., see Rodgers, J.L. 269n Oliner, S.D. 754, 771, 774n, 785n, 1291, 1294 Oliner, S.D., see Aizcorbe, A. 749 Olivei, G., see Klein, M. 907n Oliver, J. 1687 Olivetti, C., see Fernandez, R. 262n, 1268 Olley, G.S. 481 Olshen, R., see Breiman, L. 619n Olson, M.C. 429n, 434, 658 Olson, O., see Hibbs, D.A. 277n Olsson, O. 1120, 1123
I-23 Onatski, A., see Kremer, M. 596, 597, 827 Oniki, H. 1463n Ono, Y. 1425n, 1574 Oostendorp, R., see Bigsten, A. 1650 Oostendorp, R., see Freeman, R.B. 488n Opal, C., see Fay, M. 1560 Oppenheimer, V.K. 1268 Organization for Economic Co-operation and Development, 807n Ormrod, D. 1161n O’Rourke, K.H. 217 O’Rourke, K.H., see Findlay, R. 217 Ortega, F. 1442n Ortigueira, S. 1327, 1329 Ortigueira, S., see Ladron de Guevara, A. 26 Ospina, S., see Huggett, M. 1295n Ostrom, E. 1643, 1659 Ostrom, E., see Varughese, G. 1675, 1676 Oswald, A., see Blanchflower, D. 511n, 1692n Ottaviano, G.I.P. 349, 1547, 1579 Ottaviano, G.I.P., see Baldwin, R.E. 281n, 1425n, 1486n Oulton, N., see Basu, S. 1296 Overman, H.G. 1547, 1555 Overman, H.G., see Duranton, G. 1557 Overman, H.G., see Ioannides, Y.M. 1548, 1553, 1554 Overton, M. 435, 436 Owen, A., see Judson, R.A. 634 Owen, P., see Knowles, S. 653 Owens, T., see Barr, A. 1657 Oxley, L., see Greasley, D. 601 Ozler, S., see Alesina, A. 652, 655, 656, 662 Paap, R. 594, 595 Paasch, K., see Teachman, J. 1676, 1679 Pagan, A., see Breusch, T. 644 Pagano, M. 923 Pakes, A., see Olley, G.S. 481 Paldam, M. 1659 Palloni, A. 1673, 1675, 1680 Panagariya, A., see Lopez, R. 985n Pandey, P., see Hoff, K. 330 Pannabecker, J.R. 1137n Papageorgiou, C. 619, 620n Papageorgiou, C., see Duffy, J. 257n, 605n, 620, 938 Papageorgiou, C., see Masanjala, W. 612–615 Papanek, G. 1733n Papell, D.H., see Ben-David, D. 1091n Papell, D.H., see Li, Q. 602
I-24 Papell, D.H., see Loewy, M. 602 Parcel, T. 1679 Parente, S.L. 136n, 277, 303, 434, 507, 715, 820, 821, 827, 831, 835, 836n, 837, 843, 844, 943n, 948n, 1263, 1375, 1376, 1380, 1390, 1392, 1392n, 1395, 1395n, 1396, 1404, 1405n, 1406, 1407, 1413n, 1414 Parente, S.L., see Gollin, D. 721, 723n Pargal, S. 1675 Parker, R.P. 754 Parker, R.P., see Landefeld, J.S. 750n Parry, J.H. 453 Paserman, M., see Gould, E.D. 1359 Paterson, D., see Cain, L. 1311 Patillo, C., see Bigsten, A. 1650 Patrick, H.T. 875 Patrick, H.T., see Cameron, R. 909 Patrinos, H.A. 1035 Patrinos, H.A., see Psacharopoulos, G. 206n, 484, 487, 489, 834, 1035n, 1727–1729 Paukert, F. 1712, 1732 Paukert, F., see Lecaillon, J. 1732 Paulson, A. 542 Paxson, C., see Case, A. 518n Paxson, C., see Deaton, A. 544n Paxton, P. 1683, 1685, 1687 Pearlman, J., see Chui, M. 1425n Peden, E.A. 1741n Peek, J. 884n Pennacchi, G., see Gorton, G. 878n Pennington, D.H., see Brunton, D. 455 Pereira, A.S. 255 Peretto, P.F. 93n, 1094 Peretto, P.F., see Connolly, M. 260n Perez, S., see Hoover, K. 611, 612, 643 Perez-Sebastian, F., see Duffy, J. 257n Perkins, W.R., see Amir, R. 350 Perotti, R. 398, 1609 Perotti, R., see Alesina, A. 398 Perri, F., see Heathcote, J. 1425n Perri, F., see Krueger, D. 1355–1357 Perron, P. 601 Persson, J. 586n Persson, T. 398, 399, 538n, 636n, 642, 655, 1600n Pesaran, M.H. 584, 601, 634, 644 Pesaran, M.H., see Binder, M. 581, 592, 645 Pesaran, M.H., see Lee, K.S. 586n, 629, 634, 635 Pessoa, S. 483 Petersen, B., see Fazzari, S. 511n
Author Index Petersen, M.A. 889 Peterson, S.R., see Jaffe, A.B. 1763n, 1798 Pettinato, S., see Graham, C. 1692n Petzold, C. 747n Pfenning, W., see Flora, P. 205–207, 209, 211, 215, 226, 255 Pfleiderer, P., see Bhattacharya, S. 871 Phelps, E.S. 39, 44, 1070 Phelps, E.S., see Nelson, R.R. 78, 98, 102, 239, 270n, 939, 1205, 1301, 1304 Philippe, M., see Baldwin, R.E. 281n Philippon, T., see Chaney, T. 1342 Philipson, T.J., see Becker, G.S. 852n Phillips, P. 635, 644 Philpin, C.H.E., see Aston, T.H. 441 Picon, A. 1136n Pierce, B., see Juhn, C. 1285, 1286 Piketty, T. 92n, 354n, 424, 537, 539n, 1283, 1597n, 1604n, 1720 Piketty, T., see Aghion, P. 82n Pincus, S. 455 Pineda, J. 1608n, 1609, 1611 Ping, W., see Bond, E. 26 Piore, M.J. 1324n Pirenne, H. 398, 440 Pischke, J.-S., see Acemoglu, D. 1328 Pissarides, C.A. 398, 1331, 1332n, 1346 Pissarides, C.A., see Mortensen, D.T. 1331, 1332, 1346, 1349 Pistaferri, L., see Meghir, C. 1331n Pistor, K. 1007, 1008 Pistor, K., see Berkowitz, D. 1007, 1008 Platteau, J.-P. 370n, 1646, 1653, 1655, 1663n Plosser, C.I., see King, R.G. 880 Poewe, K. 1656 Polanyi, M. 1121, 1123 Polemarchakis, H.M., see Galor, O. 1425n Pollak, R., see Lundberg, S. 519 Pollard, S. 910n, 1157n, 1161 Polterovich, V. 1002n Pomeranz, K. 217, 217n, 280n, 1135, 1171n Poole, K., see Londregan, J.B. 661 Popov, V., see Polterovich, V. 1002n Popp, D. 1796 Porter, M.E. 165 Porter, R. 1127, 1135n, 1136n, 1140 Portes, A. 1643n, 1644, 1645 Portes, R. 1504n Porteus, E., see Kreps, D. 1602n Portney, P.R., see Jaffe, A.B. 1763n, 1798 Postan, M.M. 428n, 440
Author Index Postel-Vinay, G., see Piketty, T. 92n Postlewaite, A., see Cole, H. 544 Powell, C. 498 Powell, R. 429n Pratt, J., see Krasker, W. 649n Pregibon, D., see Draper, D. 582 Prescott, E.C. 136n, 314n, 316n, 605–607, 684, 1410n Prescott, E.C., see Boyd, J.H. 871 Prescott, E.C., see Cooley, T.F. 1294 Prescott, E.C., see Hansen, G.D. 176n, 236n, 262–264, 264n, 1099, 1238n, 1264, 1374–1376, 1380, 1383, 1386, 1389, 1396 Prescott, E.C., see Hornstein, A. 1393n Prescott, E.C., see Lucas Jr., R.E. 1336, 1342 Prescott, E.C., see McGrattan, E.R. 25, 1325, 1407n Prescott, E.C., see Parente, S.L. 136n, 277, 303, 434, 507, 820, 821, 827, 831, 835, 836n, 837, 843, 844, 943n, 948n, 1375, 1376, 1380, 1390, 1392, 1392n, 1395, 1395n, 1396, 1404, 1405n, 1406, 1407 Prescott, E.C., see Stokey, N.L. 311n, 320, 376 Preston, I., see Blundell, R. 1331n, 1355, 1356 Price, D.J.D. 1151n, 1152 Priestley, J. 1134n Pritchett, L. 76, 219, 256, 316n, 331, 562, 573, 625, 630, 631, 636, 637, 641, 646, 710, 716, 830, 833, 1018n, 1020n, 1441, 1515n Pritchett, L., see Easterly, W. 326, 565, 568n, 625, 655, 656, 659, 829, 996, 1036n Pritchett, L., see Hausmann, R. 646, 990, 992 Pritchett, L., see Lindauer, D.L. 973n, 1009 Pritchett, L., see Narayan, D. 1666, 1673, 1675 Prskawetz, A., see Kogel, T. 236n, 262 Przeworski, A. 277n Psacharopoulos, G. 206n, 484–487, 489, 686, 694, 834, 1035n, 1727–1729 Puga, D. 1425n, 1487n, 1578 Puga, D., see Duranton, G. 1547, 1564, 1565, 1575 Putnam, R.D. 658, 1643, 1648, 1685, 1693, 1693n Putnam, R.D., see Helliwell, J. 402, 658, 1020n, 1664, 1682, 1689 Putterman, L., see Bockstette, V. 662 Qian, J., see Allen, F. 909 Qian, M., see Allen, F. 909 Qian, N. 497 Qian, Y. 980
I-25 Qian, Y., see Lau, L.J. 980 Qiu, L.D., see Lai, E.L.C. 124 Quadrini, V. 542 Quah, D.T. 77n, 84, 256, 331–333, 354n, 565, 587n, 596, 596n, 597–599, 606, 1069n, 1088n, 1549 Quah, D.T., see Durlauf, S.N. 256, 582n, 631, 804n, 1680n, 1681, 1810 Quiggin, J., see Dowrick, S. 591 Raboy, D.G., see Wiggins, S.N. 1199 Racine, J., see Maasoumi, E. 595 Radlet, S. 298 Rahn, W. 1687, 1688 Rahn, W., see Brehm, J. 1685, 1686 Rajan, R.G. 882, 883, 885, 887, 910, 912, 912n, 923, 1320n Rajan, R.G., see Diamond, D.W. 878n Rajan, R.G., see Kashyap, A. 878n Rajan, R.G., see Kumar, K.B. 917n Rajan, R.G., see Petersen, M.A. 889 Ram, R. 653 Ramakrishnan, R.T.S. 871 Ramey, G. 33, 34, 45, 150, 659 Ramey, G., see Den Haan, W. 1346 Ramey, V., see Ramey, G. 33, 34, 45, 150, 659 Ramirez, M., see Esfahani, H. 655, 662 Ramsey, F.P. xiv, 1230, 1251 Ranciere, R., see Loayza, N.V. 661, 903 Randall, A.J. 434, 1168 Randers, J., see Meadows, D.H. 1752n Rangel, A., see Helpman, E. 1337 Ranis, G., see Fei, J.C.H. 1713 Rannis, G. 1577 Rao, P.D.S. 720 Rappaport, J. 1558 Rappaport, N.J., see Berndt, E.R. 1196n Rashad, R. 752n Rasul, I., see Bandiera, O. 339, 349 Rauch, J.E. 502, 1425n, 1551, 1653, 1660, 1714 Ravallion, M. 513, 1035, 1734n, 1736n Ravallion, M., see Chen, S. 297n Ravallion, M., see Coate, S.T. 513 Ravallion, M., see Jalan, J. 339, 1035 Ravikumar, B., see Glomm, G. 1603n Ray, D. 330, 354n, 398, 513, 522n, 1577 Ray, D., see Adsera, A. 328 Ray, D., see Dasgupta, P. 373, 537 Ray, D., see Mookherjee, D. 259n, 303, 317, 326, 326n, 354n, 537, 537n, 1604n
I-26 Ray, R., see Chakraborty, S. 882 Razin, A. 224n, 1261, 1264 Razin, A., see Frenkel, J. 1425n Razo, A., see Haber, S.H. 887, 923 Rebelo, S.T. 30, 118n, 820, 823, 1017, 1018n, 1032, 1033, 1103 Rebelo, S.T., see Easterly, W. 1033 Rebelo, S.T., see King, R.G. 314n, 316n, 781n Rebelo, S.T., see Kongsamut, P. 1263 Rebelo, S.T., see Stokey, N. 24 Redding, S. 349n Redding, S., see Aghion, P. 986n Redding, S., see Griffith, R. 78n Redding, S., see Overman, H.G. 1547, 1555 Redondo-Vega, F. 1659 Rees, A. 784 Reffett, K., see Mirman, L.J. 377 Reichlin, L. 596 Reid, M.G. 1262, 1263 Reiter, S. 1120 Reitschuler, G., see Guaresma, J. 654, 661 Renaud, B. 1559, 1571 Renelt, D., see Levine, R. 493, 575, 609, 610, 645, 647, 648, 654–657, 659, 901, 1033 Resnick, D., see Zhang, X. 1706n Resnick, M., see Rosen, K. 1553 Restuccia, D. 493, 720, 721, 723n Rey, H., see Martin, P. 156, 1490n Rey, H., see Portes, R. 1504n Rey, P., see Aghion, P. 874 Reynolds, A. 431, 432 Ricardo, D. 1374 Ricci, L., see Edison, H. 907n Rice, E., see DeAngelo, H. 882 Richard, J.-F., see Berkowitz, D. 1007, 1008 Richard, S., see Meltzer, A. 1741 Richards, F.J. 947n Richardson, P.J., see Boyd, R. 266n Riker, W.H. 1532 Ringer, F. 209, 460 Rioja, F. 903 Ríos-Rull, J.-V. 1263 Ríos-Rull, J.-V., see Krusell, P. 131n, 398, 399, 434, 1196n, 1205, 1298, 1304, 1414, 1625n Ríos-Rull, J.-V., see Parente, S.L. 1413n, 1414 Rivera, B. 662 Rivera-Batiz, L.A. 120n, 124, 125, 126n, 820, 822, 1075n, 1425n Rivkin, S.G., see Hanushek, E.A. 702n Rob, R. 1659 Rob, R., see Jovanovic, B. 500n, 715
Author Index Robbins, D. 1308 Robert-Nicaud, F., see Baldwin, R. 1486n Roberts, J., see Milgrom, P. 1310, 1320 Robertson, D. 634 Robertson, D., see Hall, S. 584 Robertson, D., see Knowles, K. 1158 Robertson, R., see Toya, H. 660 Robin, J.-M., see Bowlus, A.J. 1354, 1356, 1357 Robinson, E., see Musson, A.E. 1128n, 1132 Robinson, J.A. 422, 867 Robinson, J.A., see Acemoglu, D. 193n, 277, 278, 302, 393, 397n, 402, 410, 413, 414, 417, 417n, 419, 432, 435, 448, 450, 453, 458, 461, 463, 464, 506, 639n, 651, 654, 656–658, 662, 923, 1005, 1033, 1054, 1055, 1608n, 1741 Robinson, J.A., see Baland, J.M. 1608n Robinson, J.A., see Bianchi, S. 1685, 1686 Robinson, J.A., see Duflo, E. 482, 495, 516, 521, 522 Robinson, S. 1713 Robinson, S., see Zhang, X. 1706n Robison, L. 1683, 1684n Robson, A.J. 266n, 276n Roche, D. 1136n Rodgers, J.L. 269n Rodgers, J.L., see Kohler, H. 269n Rodney, W. 428 Rodríguez, F. 558, 971n, 986, 1050, 1422n, 1609 Rodríguez, F., see Pineda, J. 1608n, 1609, 1611 Rodríguez, F., see Rodrik, D. 1515, 1516 Rodriguez, M., see Diamond, P. 733n Rodriguez, R. 282n Rodríguez-Clare, A. 143n, 349, 715, 1001 Rodríguez-Clare, A., see Klenow, P.J. 136n, 499, 604–606, 627, 646n, 647, 648, 684, 688, 689, 698, 706, 833, 834, 938, 948, 1018n, 1032, 1390n, 1445, 1466n Rodriguez-Mora, J.V., see Hassler, J. 239n, 1597n, 1612n, 1617n Rodrik, D. 278, 349, 462, 646, 654, 658, 659, 975, 981–984, 986n, 990, 996n, 997, 999, 1001, 1002, 1004n, 1005, 1006, 1515, 1516 Rodrik, D., see Alesina, A. 398, 399, 538n, 655, 1600n Rodrik, D., see Diwan, I. 130 Rodrik, D., see Hausmann, R. 646, 990, 992, 999, 1000, 1003 Rodrik, D., see Mukand, S. 1007, 1008, 1008n
Author Index Rodrik, D., see Rodríguez, F. 558, 971n, 986, 1050, 1422n Rodrik, D., see Rodriguez, R. 282n Roe, M. 873n, 923 Roemer, J. 1662 Roemer, J., see Lee, W. 1609n Rogers, C.A., see Martin, P. 36 Rogers, M., see Dowrick, S. 604n Rogerson, R. 1264, 1289, 1352 Rogerson, R., see Benhabib, J. 1263 Rogerson, R., see Fernandez, R. 259n, 1603n, 1606 Rogerson, R., see Gollin, D. 721, 723n Rogerson, R., see Parente, S.L. 1263 Rogoff, K., see Gertler, M. 1490n Rogoff, K., see Obstfeld, M. 1504n Roland, G., see Bolton, P. 1504n, 1506, 1510n Roland, G., see Dewatripont, M. 989n Roland, G., see Lau, L.J. 980 Romer, D. 424, 592 Romer, D., see Frankel, J.A. 99, 639, 659, 663, 1032, 1096, 1422n, 1515, 1525 Romer, D., see Mankiw, N.G. xii, 77, 477, 502, 578, 586n, 587n, 598, 604, 605, 618, 629, 642, 647, 657, 684, 689, 804, 805, 821, 1018n, 1022n, 1097, 1445, 1466n Romer, P.M. xin, xiin, 93n, 114, 116, 116n, 125n, 147, 155n, 166, 223, 300, 317, 318n, 388, 558, 590, 591, 606n, 659, 660, 804, 805, 820, 822, 823, 833, 838, 857, 943, 1017, 1020n, 1065, 1066, 1070, 1073, 1082, 1088–1090, 1092, 1103, 1103n, 1106, 1107, 1504, 1517n, 1618n, 1718, 1800, 1801 Romer, P.M., see Evans, G. 158, 162, 165 Romer, P.M., see Rivera-Batiz, L.A. 120n, 124, 125, 126n, 820, 822, 1075n, 1425n Roodman, D., see Easterly, W. 1050n Root, H. 1162 Rose, A. 1505n Rosen, H., see Holtz-Eakin, D. 632 Rosen, K. 1553 Rosen, L., see Geertz, C. 1656 Rosen, S. 1321 Rosenberg, N. 144, 1152, 1160, 1169 Rosengren, E.S., see Peek, J. 884n Rosenstein-Rodan, P.N. 340, 341, 398, 503, 1001 Rosenstone, S. 1608 Rosenthal, J.-L., see Piketty, T. 92n Rosenthal, S. 1547
I-27 Rosenzweig, M.R. 239n, 496, 497, 511, 514, 515, 1722n Rosenzweig, M.R., see Behrman, J.R. 709 Rosenzweig, M.R., see Binswanger, H. 508 Rosenzweig, M.R., see Foster, A.D. 239n, 482, 495–497, 500n, 516, 938 Rosenzweig, M.R., see Munshi, K. 497 Ross, J.I. 450n Rossi, P.E., see Jones, L.E. 23, 23n Rossi-Hansberg, E. 1554, 1564–1566, 1573, 1576 Rossi-Hansberg, E., see Garicano, L. 1321 Rossi-Hansberg, E., see Lucas, R.E. 1565 Rostow, W.W. 304n, 330, 719n, 971 Rotella, E., see Cain, L. 1163n Roubini, N. 653, 923 Roubini, N., see Alesina, A. 652, 655, 656, 662 Rousseau, P.L. 653, 661–663, 903–906, 1200n Rousseau, P.L., see Jovanovic, B. 525, 1185, 1189, 1204, 1206, 1207, 1210, 1296, 1335n, 1344 Routledge, B. 1658 Rowe, D.C., see Rodgers, J.L. 269n Roy, D., see Subramanian, A. 984 Roy, S., see de Hek, P.A. 39 Rubin, R., see Somanathan, E. 1659 Rubinstein, Y. 1721 Rudolph, T., see Rahn, W. 1687, 1688 Ruelle, D. 1145n Ruffin, R.J. 1425n Ruiz-Arranz, M. 1299n Rupasingha, A. 658 Russell, R., see Henderson, D. 605, 605n, 606, 607 Rustichini, A., see Benhabib, J. 398, 538n Rutherford, T.F., see Harrison, G.W. 985n Ruttan, V.W. 748n Sabel, C.F., see Piore, M.J. 1324n Sabot, R.H., see Gelb, A. 711n Sacerdote, B., see Alesina, A. 1597n, 1609 Sacerdote, B., see Glaeser, E.L. 1020n, 1667 Sachs, J.D. 298, 300n, 304n, 400, 413, 615n, 653–657, 659, 954, 956, 956n, 1050, 1422n, 1515, 1515n Sachs, J.D., see Bloom, D.E. 400, 652, 654, 655, 657 Sachs, J.D., see Gallup, J.L. 655, 662, 1560 Sachs, J.D., see Hamoudi, A. 655 Sachs, J.D., see Lipton, D. 989 Sachs, J.D., see Masters, W.E. 654, 658 Sachs, J.D., see McArthur, J.W. 654, 656
I-28 Sacks, D., see Rappaport, J. 1558 Sadik, J., see Berdugo, B. 279n Sadoulet, E., see de Janvry, A. 1714 Saenz, L., see Lederman, D. 833, 834, 846, 849 Saez, E., see Piketty, T. 1283 Saggi, K., see Glass, A.J. 1425n Sahasakul, C., see Barro, R.M. 24 Sahay, R., see Cashin, P. 586n Saint-Paul, G. 266, 398, 876, 1321, 1597n, 1603n, 1742n Saint-Paul, G., see Bentolila, S. 1340 Saito, M., see Devereux, M.B. 1425n Sakellaris, P. 54n, 1294 Sakellaris, P., see Dellas, H. 54n Sala-i-Martin, X. 219, 493, 586n, 611, 612, 650n, 653–659, 1032, 1097 Sala-i-Martin, X., see Barro, R.J. xin, 16n, 28, 77, 78n, 120n, 122n, 139n, 196n, 495, 582n, 586n, 587, 587n, 591, 604n, 605, 606, 647, 660, 821, 827, 835, 837–839, 938, 940, 944, 945n, 1046, 1099n, 1103n, 1509n, 1523, 1560 Sala-i-Martin, X., see Doppelhofer, G. 587, 612–615 Sala-i-Martin, X., see Mulligan, C.B. 1103n Sala-i-Martin, X., see Roubini, N. 653, 923 Samaniego, R.M. 1414 Sampson, R. 1687, 1688 Samuelson, P.A. 131n, 734n, 1446 Samuelson, P.A., see Dornbusch, R. 143, 1463n, 1468 Sandberg, L.G. 1155n Sandefur, G. 1677, 1679 Sandefur, R. 1690n Sanderson, M. 197, 207, 207n, 208 Santos, M.S., see Ladron de Guevara, A. 26 Sapienza, P., see Guiso, L. 896, 896n, 908, 923, 1678, 1682, 1684 Sarel, M. 1044n Sargent, R.-M. 1135 Sargent, T.J., see Ljungqvist, L. 1342, 1344, 1346 Sarte, P., see Li, W. 26 Sattinger, M. 1338 Satyanath, S., see Miguel, E. 660 Sauré, P. 1425n Savvides, A. 79n Savvides, A., see Mamuneas, T. 618 Saw, P.L.S., see Wellisz, S. 984 Sawada, Y., see Jimenez, E. 1658 Scharfstein, D. 874
Author Index Schattschneider, E.E. 434 Scheffer, M. 1760n Scheinkman, J., see Glaeser, E.L. 1553, 1689 Schelifer, A., see Glaeser, E.L. 1553 Scherer, F. 89 Schiantarelli, F., see Bond, S. 627, 660 Schiff, E. 1166 Schlicht, E. 1719n Schlozman, K., see Brady, H. 1741n Schmalensee, R. 1818n Schmidt, P., see Ahn, S. 633 Schmidt, R., see Kelley, A. 657 Schmidt-Traub, G., see Sachs, J.D. 298, 300n, 304n Schmitz, J.A. 718 Schmitz, J.A., see McGrattan, E.R. 804n, 1019n Schmitz Jr., J.A. 1392n Schmitz Jr., J.A., see Holmes, T.J. 1414 Schmitz Jr., J.A., see McGrattan, E.R. 684n Schmookler, J. 133n Schmookler, J., see Griliches, Z. 133n Schmukler, S., see Levine, R. 896n Schoar, A.S. 1206 Schoar, A.S., see Bertrand, M. 909 Schofield, R.S. 273 Schofield, R.S., see Wrigley, E.A. 182n, 184, 185, 191, 192, 204, 226, 255 Schreyer, P. 746, 756, 784, 785, 785n, 787 Schreyer, P., see Colecchia, A. 785n, 786n, 1346 Schultz, T.P. 497, 528n Schultz, T.P., see Rosenzweig, M.R. 496, 497 Schultz, T.W. 239n, 270n Schumpeter, J.A. 113, 133n, 867, 871 Schwert, G.W. 1206n Scotchmer, S. 144, 145 Scott, J.C. 434 Scrimshaw, N.S., see Basta, S.S. 490 Seers, D., see Meier, G.M. 867 Segenti, E., see Miguel, E. 660 Segerstrom, P.S. xiin, 71n, 114, 127n, 937, 1070n, 1089, 1105, 1425n Segerstrom, P.S., see Dinopoulos, E. 127n, 142, 1307, 1308n, 1425n Segerstrom, P.S., see Lundborg, P. 1442n Seki, E., see Platteau, J.-P. 1653, 1663n Selden, T. 1766n, 1767n Selden, T., see Holtz-Eakin, D. 1818n Sen, A., see Hoff, K. 368, 369 Sener, ¸ F. 1425n
Author Index Serageldin, I., see Dasgupta, P. 1648 Servén, L., see Kraay, A. 1490n Seshadri, A., see Greenwood, J. 229n, 262n, 1098n, 1194, 1227n, 1230n, 1231n, 1238n, 1245n, 1252n, 1262n, 1264n, 1312, 1312n Seshadri, A., see Manuelli, R.E. 62 Sevilla, J., see Bloom, D.E. 256, 277n, 335, 621, 654, 655 Shaban, R. 508 Shafik, N. 1766n Shan, J.Z. 892 Shapin, S. 1132n, 1133n Shapiro, C. 807n Sharfstein, D., see Hoshi, T. 884n, 918n Sharif, R. 941 Shastry, G.K. 708 Shaw, E.S., see Gurley, J.G. 867, 875 Shell, K. 17, 113, 1070 Sheshadri, A. 1603n, 1606 Sheshinski, E. 1070n Sheshinski, E., see Levhari, D. 113n Shi, S. 1339 Shi, S., see Devereux, M.B. 1463n Shibata, A., see Ono, Y. 1425n Shiller, R.J. 806n Shiller, R.J., see Campbell, J.Y. 768n, 806n Shin, Y., see Pesaran, M.H. 634 Shioji, E. 586n, 650 Shipp, S., see Johnson, D.G. 1355 Shishido, H., see Wheaton, W. 1561 Shleifer, A. 158n, 873, 873n, 882, 999, 1117 Shleifer, A., see Djankov, S. 99, 425, 426, 506, 509n, 1006–1008, 1395 Shleifer, A., see Glaeser, E.L. 278–280, 425, 426 Shleifer, A., see La Porta, R. 84n, 399, 425, 426, 506, 658, 874, 887, 892, 897, 922, 923, 1033, 1505, 1683 Shleifer, A., see Murphy, K.M. 145, 340–342, 364, 365, 398, 503, 506n, 653, 711n, 996n, 999, 1001, 1021, 1504 Sibert, A. 1425n Sichel, D.E., see Aizcorbe, A. 749 Sichel, D.E., see Oliner, S.D. 754, 771, 774n, 785n, 1291, 1294 Sicherman, N., see Bartel, A. 1301, 1318 Sickles, R., see Hultberg, P. 663 Siebert, W.S., see Addison, J.T. 1330, 1330n Siegel, D., see Lichtenberg, F.R. 1206 Siegelbaum, L. 1656
I-29 Siegler, M.V. 37 Siles, M., see Robison, L. 1683, 1684n Simhon, A., see Gould, E.D. 231 Simon, B. 208 Simon, H.A. 301, 1554 Simon, J.L. 1069, 1070, 1148n Singh, R., see Nelson, M. 654 Sirri, E.R. 879, 880 Siu, H., see Jones, L.E. 39, 60, 61, 61n Sjostrom, T., see Ghatak, M. 537 Skaperdas, S. 398 Skiba, A.K. 350, 543 Skidmore, M., see Toya, H. 660 Skocpol, T. 1649n Slade, M.E. 1751n Slemrod, J. 1033 Slok, T., see Edison, H. 907n Smeeding, T.M., see Gottschalk, P. 1350, 1733 Smith, A. 179n, 440, 880, 1117, 1157n, 1264, 1504 Smith, A., see Krussel, P. 515, 544n Smith, B.D. 922 Smith, B.D., see Bencivenga, V.R. 150n, 875, 877, 878, 895, 896, 923 Smith, B.D., see Boyd, J.H. 874, 875, 879, 886, 920, 923 Smith, B.D., see Greenwood, J. 880, 881 Smith, G.W., see Devereux, M.B. 876 Smith, J.G. 1136n, 1137n, 1141n Smith, J.G., see LaLonde, R.J. 1360 Smith, R., see Huybens, E. 886, 923 Smith, R., see Lee, K.S. 586n, 629, 634, 635 Smith, R., see Pesaran, M.H. 634 Smulders, J. 1804 Smulders, S. 1752n, 1804, 1805 Smulders, S., see Bovenberg, A.L. 1778, 1797 Smulders, S., see van de Klundert, T. 93n, 1425n Snelders, H.A.M. 1143n Snooks, G. 1117n Snower, D.J., see Lindbeck, A. 1321, 1330 Snyder, W., see Morrisson, C. 213, 214 Soares, R.R. 231n, 259 Soares, R.R., see Becker, G.S. 852n Soderbom, M., see Bigsten, A. 481, 1650 Soekirman, D.S., see Basta, S.S. 490 Sokoloff, K.L., see Engerman, S.L. 212n, 278, 302, 443, 449, 923, 1005, 1054, 1611, 1617 Sokoloff, K.L., see Khan, B.Z. 1165, 1166 Solmon, L., see Landes, W. 1249
I-30 Solow, R.M. xi, 17, 113, 223, 298, 307, 388, 558, 560, 715, 768, 779, 779n, 780n, 781–783, 783n, 784, 791, 792, 804–806, 821, 1070, 1074, 1101–1103, 1230, 1251, 1290, 1332, 1375, 1382, 1421, 1718, 1752n, 1772, 1788, 1791n Solow, R.M., see Baily, M.N. 718 Solow, R.M., see Dertouzos, M. 782n Somanathan, E. 1659 Song, D., see Selden, T. 1766n, 1767n Song, S., see Baltagi, B. 644 Soon, C. 996 Sorensen, B.E., see Kalemli-Ozcan, S. 157 Sorenson, A., see Morgan, S. 1677, 1679, 1680n Soutter, C., see Glaeser, E.L. 1689 Spadafora, D. 1136n Spagat, M., see Bertocchi, G. 259n Sparrow, W.J. 1141 Spence, M. 116, 1070, 1082 Spence, M., see Keeler, E. 1774n, 1786, 1810n Spencer, B.J., see Dowrick, S. 1409n Spiegel, M.M., see Benhabib, J. 81, 495, 903, 938, 939, 953, 959, 959n, 1018n Spilimbergo, A., see Sakellaris, P. 54n Spittel, M., see Palloni, A. 1673, 1675, 1680 Spolaore, E. 1502n, 1506, 1508n, 1509n, 1513, 1514n, 1518n, 1519, 1525 Spolaore, E., see Alesina, A. 1422n, 1502, 1502n, 1506, 1510, 1510n, 1511n, 1513, 1519, 1519n, 1525n, 1530, 1530n, 1535n Spree, R. 214 Squire, L., see Deininger, K. 538, 1732 Squire, L., see Li, H. 462 Squire, L., see Lundberg, M. 1733n Srinivasan, S., see Basu, S. 1296 Srinivasan, T.N., see Levhari, D. 39, 44, 870 Stabler, J.C., see Clark, J.S. 1553 Stacchetti, E., see Jones, L.E. 39, 60, 61, 61n Stachurski, J. 311n, 332n, 377n Stachurski, J., see Azariadis, C. 333, 398, 597 Stachurski, J., see Mirman, L.J. 377 Stachurski, J., see Nishimura, K. 308n, 350 Stafford, F.P., see Johnson, G.E. 703n Standard and Poor’s Corporation, 1210n Stangeland, D., see Morck, R. 874, 884n Stark, O., see Rosenzweig, M.R. 1722n Starr, R.M., see Bencivenga, V.R. 877, 895, 896 Starrett, D. 349 Stasavage, D. 456
Author Index Stavins, R.N., see Newell, R.G. 1796 Steel, M., see Fernandez, C. 587, 587n, 612, 613, 613n, 614, 1681 Stein, J.C. 874 Stein, J.C., see Kashyap, A. 878n Stengos, T., see Kalaitzidakis, P. 610 Stengos, T., see Liu, Z. 589, 617, 618 Stengos, T., see Maasoumi, E. 595 Stengos, T., see Mamuneas, T. 618 Stephenson, E.F., see Dawson, J.W. 37 Stern, B.J. 1168 Stern, N. 998 Stern, S., see Porter, M.E. 165 Stevenson, J. 458 Stewart, L. 1151 Stigler, G.J. 781 Stiglitz, J.E. 143n, 509, 513, 872, 881, 882, 974n, 1103, 1463n, 1567, 1715, 1717n, 1718, 1772, 1788, 1789, 1790n Stiglitz, J.E., see Atkinson, A.B. 131n Stiglitz, J.E., see Dixit, A.K. 113, 116, 1070, 1082 Stiglitz, J.E., see Grossman, S.J. 872 Stiglitz, J.E., see Hellmann, T. 981, 982 Stiglitz, J.E., see Hoff, K. 367n, 480n, 999, 1001 Stiroh, K.J. 778n, 779 Stiroh, K.J., see Jorgenson, D.W. 757n, 759n, 760, 760n, 763n, 769, 771, 774, 779, 782n, 785n, 1291 Stock, J., see Kremer, M. 596, 597, 827 Stoker, T.M., see Schmalensee, R. 1818n Stokey, N.L. 24, 187n, 217, 311n, 320, 376, 820–822, 1099, 1425n, 1755, 1756, 1770, 1772, 1778, 1779, 1784, 1784n, 1786, 1788, 1792, 1794n, 1804, 1804n, 1810, 1811 Stokey, N.L., see Alvarez, F. 41n Stokey, N.L., see Rebelo, S.T. 1033 Stolyarov, D., see Laitner, J. 1207 Stone, C., see Breiman, L. 619n Stone, L. 273 Storesletten, K., see Hassler, J. 1597n, 1617n Storesletten, K., see Heathcote, J. 1282n, 1331n, 1355–1357 Strahan, P.E., see Jayaratne, J. 907, 907n Strange, W., see Helsley, R. 1564, 1568, 1573, 1574, 1586 Strange, W., see Rosenthal, S. 1547 Strauss, J. 490, 491, 512n, 519 Streufert, P., see Ray, D. 354n Strulik, H., see Funke, M. 145n
Author Index Stulz, R.M. 873, 918, 923 Subramanian, A. 984 Subramanian, A., see Rodrik, D. 278, 654, 658, 659, 990, 1005, 1516 Sue Wing, I. 1795 Sugden, R. 329 Sul, D., see Phillips, P. 635, 644 Summerhill, W., see North, D.C. 419 Summers, L.H. 978n Summers, L.H., see DeLong, J.B. 643, 656, 660 Summers, L.H., see Easterly, W. 326, 565, 568n, 625, 655, 656, 659, 829, 996, 1036n Summers, L.H., see Krueger, A.B. 1331n Summers, L.H., see Shleifer, A. 882 Summers, R. 582, 685, 720, 804, 1037, 1403–1406 Summers, R., see Heston, A. xii, 303n, 491, 562, 685, 804n, 826, 828, 830, 831, 834, 1412, 1519 Summers, R., see Kravis, I.B. 803 Sun, F., see Shan, J.Z. 892 Sun, Y. 1679 Sunde, U., see Cervellati, M. 259, 1145 Sussman, N., see Berdugo, B. 279n Sussman, O. 875 Sussman, O., see Harrison, P. 875 Svendsen, G., see Paldam, M. 1659 Svensson, J. 506 Swagel, P., see Alesina, A. 652, 655, 656, 662 Swallow, B.M., see Barrett, C.B. 326n Swan, T.W. 113, 558 Swank, D. 659 Swartz, S. 641 Sylla, R.E. 886 Sylla, R.E., see Homer, S. 1213n, 1220n Sylla, R.E., see Rousseau, P.L. 653, 661, 906 Symons, J., see Robertson, D. 634 Syropoulos, C., see Dinopoulos, E. 1425n Syrquin, M. 1724n Syrquin, M., see Chenery, H.B. 1712, 1724n Tabellini, G., see Giavazzi, F. 636, 636n, 646 Tabellini, G., see Persson, T. 398, 399, 538n, 636n, 642, 655, 1600n Taber, C., see Heckman, J.J. 1318, 1319, 1354–1357 Tabuchi, T. 1578 Tadesse, S. 919 Tahvonen, O. 1797 Takahashi, T., see Matsuyama, K. 1487n
I-31 Takeyama, L., see Johnson, P.A. 620n Tamura, R.F. 236n, 261, 263, 820, 821, 1099, 1378, 1618n Tamura, R.F., see Becker, G.S. 224n, 235, 1021, 1378 Tan, C.M. 620 Tan, H.W., see Lillard, L.A. 1318 Tarr, D.G., see Harrison, G.W. 985n Tarrow, S. 450n, 462, 1691n Tavares, J. 638, 661 Tawney, R.H. 393, 402 Taylor, A.M. 281n Taylor, A.M., see Estevadeordal, A. 216, 1532, 1532n Taylor, C. 639n, 1651 Taylor, L., see Bacha, E.L. 1714 Taylor, L.L., see Hanushek, E.A. 702n Taylor, M.S. 1425n Taylor, M.S., see Antweiler, A. 1770n, 1771n, 1788 Taylor, M.S., see Brander, J.A. 1752n Taylor, M.S., see Brock, W.A. 1756, 1770n, 1772n, 1799, 1800, 1809, 1811n, 1815, 1816 Taylor, M.S., see Copeland, B.R. 1755, 1757n, 1773, 1779, 1786, 1788, 1799n, 1804 Taylor, M.S., see Levinson, A. 1764 Teachman, J. 1676, 1679 Teal, F., see Bigsten, A. 1650 Teixeira, A. 1414 Teixeira, A., see Herrendorf, B. 1395n Temple, J.R.W. 560n, 562n, 580, 590, 591, 611n, 626, 626n, 629–631, 641, 642, 658, 659, 725n, 938, 971n, 997, 1032, 1680n, 1681 Temple, J.R.W., see Bond, S. 632 Temple, J.R.W., see Graham, B.S. 256, 337, 607n, 623, 723n, 727 Tenreyro, S., see Alesina, A. 1505n Tenreyro, S., see Caselli, F. 690 Tenreyro, S., see Koren, M. 718n ter-Weel, B., see Borghans, J.A.M. 266 Terrell, K. 1730 Thaicharoen, Y., see Acemoglu, D. 464, 662, 1055 Thakor, A., see Boot, A.W.A. 881–883 Thakor, A., see Ramakrishnan, R.T.S. 871 The Commercial and Financial Chronicle, 1200n, 1206n The New York Times Co., 1200n Thesmar, D. 1324, 1619n Thesmar, D., see Bertrand, M. 909 Thisse, J.-F., see Fujita, M.P. 1579
I-32 Thisse, J.-F., see Ottaviano, G.I.P. 349, 1547, 1579 Thoenig, M. 142n, 1627n Thoenig, M., see Thesmar, D. 1324, 1619n Thomas, D. 490 Thomas, D., see Currie, J. 704, 705n Thomas, D., see Strauss, J. 490, 491, 512n, 519 Thomas, R.P., see North, D.C. 150, 388, 423, 427, 463, 1005 Thomas, V. 1570, 1724n Thomis, M.I. 434 Thompson, P., see Dinopoulos, E. 93n, 1094 Thorne, S. 1151n Thoursie, P., see Agell, J. 662 Thursby, M., see Jensen, R. 1425n Tilly, C. 462 Tilly, R., see Cameron, R. 909 Timberg, T. 479, 480, 483, 525 Timmer, M., see van Ark, B. 786n Tinbergen, J. 781, 781n, 784, 1721, 1724n, 1729 Tiongson, E., see de Mello, L. 1609 Tirole, J. 69, 367, 518 Tirole, J., see Bénabou, R. 1597n Tirole, J., see Holmstrom, B. 872, 878, 895 Tobin, J. 768 Todaro, M., see Harris, J.R. 725n, 1577 Tollison, R.D., see Ekelund Jr., R.B. 1162 Tomes, N., see Becker, G.S. 1261, 1262, 1264 Topa, G., see Conley, T.G. 645 Topel, R., see Murphy, K.M. 1288n, 1359 Tornell, A. 398, 464 Townsend, R.M. 423, 513, 542, 543, 874, 1739n Townsend, R.M., see Jeong, H. 543 Townsend, R.M., see Paulson, A. 542 Toya, H. 660 Traber, E., see Heckman, J.J. 1739n Trajtenberg, M., see Bresnahan, T.F. 1185, 1295 Trajtenberg, M., see Helpman, E. 807n, 1200, 1207 Tranter, N.L. 1148 Travis, A. 1152 Treadway, R., see Coale, A.J. 190, 200n Trebbi, F., see Rodrik, D. 278, 654, 658, 659, 1005, 1516 Trefler, D. 1466 Tressel, T. 1742n Trindade, V. 1001 Trindade, V., see Rauch, J.E. 1653
Author Index Triplett, J.E. 750n, 751, 774, 774n, 799, 806n Triplett, J.E., see Bosworth, B.P. 746n, 771n Troeskan, W., see Beeson, P.E. 1558 Troske, K.R., see Doms, M. 270n Tsiddon, D. 351n Tsiddon, D., see Brezis, E.S. 278n Tsiddon, D., see Brezis, S.L. 1427n, 1453n Tsiddon, D., see Dahan, M. 224n Tsiddon, D., see Galor, O. 239n, 1302, 1613n, 1721n Tsiddon, D., see Rubinstein, Y. 1721 Tsionas, E.G., see Christopoulos, D.K. 906 Tufano, P., see Sirri, E.R. 879, 880 Tukey, J., see Friedman, J. 620n Tullock, G., see Grier, K.B. 33, 581n Turnovsky, S.J. 56n U.S. Bureau of the Census, 233, 1227n, 1258n Udry, C.R. 480, 513, 519 Udry, C.R., see Bardhan, P. 398, 513 Udry, C.R., see Conley, T.G. 339, 349, 516 Udry, C.R., see Goldstein, M. 482, 495, 508 Ueda, K., see Townsend, R.M. 542, 543, 1739n Uglow, J. 1135n, 1137, 1162n Ulph, D., see Beath, J. 349 Unger, R.M. 1007n United Nations, 784n United Nations Development Program, 491n United States Bureau of Economic Analysis, 1188n, 1191n, 1200n, 1201n, 1218n, 1220n United States Bureau of the Census, Department of Commerce, 1184n, 1193n, 1196n, 1198n, 1218n University of Chicago Center for Research on Securities Prices, 1200n Uphoff, N., see Krishna, A. 1674 Urrutia, C., see Restuccia, D. 493 Uzawa, H. 23 Uzawa, H., see Oniki, H. 1463n Valev, N., see Rioja, F. 903 Valls Pereira, P., see Andrade, E. 598 Vamvakidis, A. 282n, 986 Van, P.H., see Basu, K. 373 van Ark, B. 786n van Ark, B., see Wagner, K. 718n Van Arkadie, B. 976n van Bastelear, T., see Grootaert, C. 1648 van de Klundert, T. 93n, 1425n van der Woude, A.M., see de Vries, J. 453, 454, 1161n
Author Index van Dijk, H., see Paap, R. 594, 595 van Elkan, R. 1425n Van Huyck, J.B. 330 Van Reenen, J., see Blundell, R. 76, 86, 89 Van Reenen, J., see Caroli, E. 1322, 1619n Van Reenen, J., see Griffith, R. 78n Van Reenen, J., see Hall, B.H. 847 van Schalk, T., see Beugelsdijk, S. 658, 1682 Van Zanden, J.L. 1158 Vandenbroucke, G., see Greenwood, J. 1227n, 1231n, 1262n Vandenbussche, J., see Aghion, P. 102, 102n, 104 Vanek, J. 1425n Varangis, P., see Gilbert, C.L. 985n Vargas Llosa, M. 414 Varian, H.R., see Shapiro, C. 807n Varoudakis, A., see Berthelemy, J. 653 Varughese, G. 1675, 1676 Veitch, J.M. 394 Velasco, A., see Hausmann, R. 1003 Velasco, A., see Tornell, A. 398 Véliz, C. 402, 419 Velloso, F., see Pessoa, S. 483 Venables, A.J. 1485n Venables, A.J., see Fujita, J. 1578, 1579 Venables, A.J., see Fujita, M.P. 1021, 1486n Venables, A.J., see Henderson, J.V. 1585 Venables, A.J., see Krugman, P.R. 143n, 398, 1485 Venables, A.J., see Limao, N. 300n, 349n Venables, A.J., see Overman, H.G. 1547, 1555 Venables, A.J., see Puga, D. 1425n Venables, A.J., see Redding, S. 349n Ventura, J. 143n, 733, 1430n, 1452n, 1453n, 1464n, 1490n Ventura, J., see Acemoglu, D. 820, 835, 1453n, 1472n Ventura, J., see Broner, F. 1490n Ventura, J., see Kraay, A. 1490n Verba, S., see Brady, H. 1741n Verdier, T., see Ades, A.F. 398 Verdier, T., see Bisin, A. 266n Verdier, T., see Bourguignon, F. 259n, 399, 1608n, 1741n Verdier, T., see Saint-Paul, G. 398, 1603n Verdier, T., see Thoenig, M. 142n, 1627n Vermeersch, C. 498 Vernon, R. 127 Verrecchia, R.E., see Diamond, D.W. 874 Vickers, B. 1134
I-33 Vickers, J., see Aghion, P. 86, 89, 398, 938 Villanueva, D., see Knight, M. 579, 683, 1446n Villanueva, D., see Loayza, N.V. 1425n Vincenti, W. 1153 Vindigni, A. 1598n, 1623 Vines, D., see Molana, H. 1425n Violante, G.L. 1286, 1316, 1333n, 1334, 1335, 1337, 1721 Violante, G.L., see Acemoglu, D. 1327, 1597n, 1617 Violante, G.L., see Aghion, P. 102n, 1302, 1313, 1314n, 1315, 1333, 1721 Violante, G.L., see Cummins, J.G. 1189, 1292, 1293, 1295, 1304, 1305, 1309, 1309n, 1337 Violante, G.L., see Heathcote, J. 1282n, 1331n, 1355–1357 Violante, G.L., see Hornstein, A. 1332, 1333, 1346, 1347, 1347n, 1349, 1350 Violante, G.L., see Krusell, P. 131n, 1196n, 1205, 1298, 1304 Vishny, R.W., see La Porta, R. 84n, 399, 425, 426, 506, 658, 874, 887, 897, 922, 923, 1033, 1505, 1683 Vishny, R.W., see Murphy, K.M. 145, 340–342, 364, 365, 398, 503, 506n, 653, 711n, 996n, 999, 1001, 1021, 1504 Vishny, R.W., see Shleifer, A. 873, 873n, 882, 999, 1117 Vogan, C.R. 1761, 1763 Voigt, S., see Feld, L. 656 Vollrath, D., see Galor, O. 263, 264, 279 Volpin, P., see Pagano, M. 923 von Amsburg, J., see Routledge, B. 1658 Voth, H.-J. 197, 255 Voth, J., see Antrás, P. 1130 Vroman, S., see Albrecht, J. 1338n Wachtel, P., see Rousseau, P.L. 661, 662, 903–905 Wacziarg, R. 575, 630, 636, 637, 646, 647, 659, 1515, 1516, 1516n Wacziarg, R., see Alesina, A. 278, 653, 1422n, 1502, 1503, 1503n, 1505, 1506, 1510n, 1514n, 1518n, 1519, 1519n, 1525n, 1535n, 1538n Wacziarg, R., see Imbs, J. 157, 1000n Wacziarg, R., see Spolaore, E. 1506, 1508n, 1509n, 1513, 1514n, 1518n, 1519, 1525 Wacziarg, R., see Tavares, J. 638, 661 Wade, R. 1653 Wagner, A. 1741
I-34 Wagner, K. 718n Walker, S., see Powell, C. 498 Wallerstein, I.M. 428 Wang, H.G., see Henderson, J.V. 1579, 1582 Wang, J.-Y. 1425n Wang, M., see Blonigen, B. 654 Wang, P., see Abdel-Rahman, H. 1574 Wang, Y., see Thomas, V. 1724n Warner, A.M. 636 Warner, A.M., see Sachs, J.D. 615n, 653–657, 659, 954, 956, 956n, 1050, 1422n, 1515, 1515n Watson, J., see Rauch, J. 1660 Wattenberg, B.J., see Caplow, T. 1251 Watts, G.W., see Cole, A.H. 1137n Weatherill, L. 1118n Weber, M. 401, 419 Weeks, M., see Corrado, L. 601n Weel, B., see Borghans, L. 1303 Wei, S.-J. 989n Weil, D.N. 200n, 708 Weil, D.N., see Basu, S. 734n, 938, 942, 943 Weil, D.N., see Galor, O. 176n, 177, 228, 229, 231n, 232, 235n, 236, 237, 255n, 257, 258, 261–263, 267n, 270n, 272n, 278, 280, 1098n, 1099, 1118n, 1146, 1261, 1264, 1266, 1267, 1378 Weil, D.N., see Mankiw, N.G. xii, 77, 477, 502, 578, 586n, 587n, 598, 604, 605, 618, 629, 642, 647, 657, 684, 689, 804, 805, 821, 1018n, 1022n, 1097, 1445, 1466n Weil, D.N., see Shastry, G.K. 708 Weil, P., see Basu, S. 1427n Weinberg, B.A. 1285, 1312, 1316, 1317 Weinberg, B.A., see Gould, E.D. 1336, 1337 Weingast, B.R. 429n Weingast, B.R., see de Figueiredo, R. 1505n Weingast, B.R., see Greif, A. 298n Weingast, B.R., see North, D.C. 419, 453, 457 Weinstein, D.E. 884, 918n Weisdorf, J.L. 259, 261 Weisenfeld, S.L. 266n Weiss, A., see Stiglitz, J.E. 872 Weitzman, M.L. 1066, 1080, 1107 Welch, F. 938, 1730, 1731 Welch, F., see Murphy, K.M. 1285 Welch, I., see Bikhchandani, S. 516n Welch, K.H., see Wacziarg, R. 636, 646, 659, 1516 Welling, L., see King, I. 1332n Wellisz, S. 984
Author Index Welsch, H. 652 Welsch, R., see Swartz, S. 641 Wenger, E. 884, 885, 918n Werner, A., see Gelos, R.G. 507, 508 West, K., see Brock, W. 559n, 610, 611n, 612–615 Western, B., see Farber, H.S. 1326 Westphal, L.E., see Evenson, R.E. 78n Wheaton, B., see Hagan, J. 1678 Wheaton, W. 1561 Whelan, K. 783n, 1106 Whinston, A.B., see Choi, S.-Y. 751n, 807n White, H. 643 White, H., see MacKinnon, J. 643 Whiteley, P. 658 Wiarda, H.J. 419 Wickens, M., see Hall, S. 584 Wiggins, S.N. 1199 Wilcoxen, P.W., see Jorgenson, D.W. 1786n Willett, J.B., see Murnane, R.J. 704, 704n Williams, D., see Kenny, C. 1032 Williams, E.E. 428 Williams, J.C. 1091n Williams, J.C., see Jones, C.I. 838, 1087 Williamson, J.G. 213, 973, 989, 1033, 1205, 1560 Williamson, J.G., see Becker, G. 1578 Williamson, J.G., see Blattman, C. 656 Williamson, J.G., see Bloom, D.E. 261n, 655 Williamson, J.G., see Clemens, M.A. 282n Williamson, J.G., see Kelly, A.C. 1578 Williamson, J.G., see Kuczynski, P.-P. 974n Williamson, J.G., see Lindert, P.H. 214 Williamson, J.G., see O’Rourke, K.H. 217 Williamson, O. 422 Williamson, R., see Stulz, R.M. 923 Williamson, S.D. 880 Williamson, S.D., see Greenwood, J. 1425n Wilson, C.H. 1161n, 1470n, 1531 Wilson, D.J. 712 Wilson, D.J., see Caselli, F. 711, 712n, 714, 714n Wilson, D.J., see Sakellaris, P. 1294 Wilson, D.M., see Harbaugh, W. 1767n Wilson, E.O., see MacArthur, R.H. 231, 269 Wilson, J.W. 1207n Winter, S., see Nelson, R.R. 1122n Wittman, D. 423 Wodon, Q., see Ravallion, M. 1035 Wolf, H., see McCarthy, D. 654 Wolfe, T. 747n
Author Index Wolfenzon, D., see Morck, R. 873, 874, 883, 887 Wolfers, J., see Blanchard, O. 1288, 1340 Wolff, E.N., see Baumol, W.J. 1715n Wolpin, K.I., see Eckstein, Z. 227n Wolpin, K.I., see Lee, D. 1299n Wolpin, K.I., see Rosenzweig, M.R. 239n, 511, 514 Wolthuis, J. 212 Wong, L.Y. 1338, 1339n, 1350n Wong, R.B. 1171n Wood, A. 140 Wood, H.T. 1142n Woodruff, C., see Johnson, S. 508, 914, 999, 1653 Woodruff, C., see McKenzie, D. 481 Woodruff, C., see McMillan, J. 1653 Woolcock, M. 1020n, 1648 World Bank, 974n, 984, 1017, 1037, 1560, 1706n, 1730n World Development Indicators, 200, 203 Wright, E.M., see Rossi-Hansberg, E. 1554, 1564, 1573, 1576 Wright, R.E. 880n, 881n, 908 Wright, R.E., see Benhabib, J. 1263 Wright, R.E., see Parente, S.L. 1263 Wright, R.E., see Williamson, S.D. 880 Wrigley, E.A. 182n, 184, 185, 191, 192, 201, 204, 226, 255 Wu, S., see Maddala, G. 618 Wu, Y., see McCarthy, D. 654 Wuketits, F. 1151 Wulf, J., see Rajan, R. 1320n Wurgler, J. 914 Wyckoff, A.W. 756n, 785, 785n, 803 Xie, D. 145n Xie, D., see Kongsamut, P. 1263 Xiong, K. 1570 Xiong, K., see Anas, A. 1570 Xu, L.C., see Clarke, G. 921n Xu, L.C., see Cull, R. 909 Xu, Y., see Campbell, J.Y. 1342 Xu, Z. 906 Yafeh, Y., see Weinstein, D.E. 884, 918n Yan, H., see Li, R. 490 Yanagawa, N. 1425n Yang, D.T., see Restuccia, D. 720, 721, 723n Yang, G. 130, 1425n Yang, S., see Brynjolfsson, E. 806n
I-35 Yanikkaya, H. 986 Yasumoto, J., see Frank, K. 1678 Yates, P.L. 217 Yeung, B., see Morck, R. 872–874, 883, 884n, 887 Yi, K.-M., see Kocherlakota, N. 626 Yin, W., see Ashraf, N. 522 Yip, C.K., see Bond, E. 26 Yip, E., see Jorgenson, D.W. 787, 788n, 805 York, A., see Clark, R. 1725n Yorukoglu, M., see Greenwood, J. 262n, 1098n, 1194, 1204, 1230n, 1252n, 1296, 1301, 1302, 1312 Yosha, O., see Kalemli-Ozcan, S. 157 Young, A.A. 93, 114, 121n, 143n, 145, 145n–147n, 153, 157, 281n, 349, 493, 494, 751, 837, 993, 997n, 1018n, 1094, 1119n, 1412n, 1425n, 1517 Young, H.P., see Durlauf, S.N. 1690 Youngson, A.J. 1169 Ypma, G., see van Ark, B. 786n Yu, W., see Morck, R. 872 Yuki, K., see Sheshadri, A. 1603n, 1606 Yun, K.-Y., see Jorgenson, D.W. 759n, 760n Zachariadis, M. 94n Zachariadis, M., see Savvides, A. 79n Zaffaroni, P., see Michelacci, C. 602, 603 Zagha, R., see Williamson, J.G. 989 Zak, P. 658, 1658, 1683 Zamarripa, G., see La Porta, R. 885 Zanforlin, L., see Calderón, C. 662 Zeckhauser, R., see Keeler, E. 1774n, 1786, 1810n Zeira, J., see Galor, O. 92n, 259n, 351, 354n, 398, 441, 537, 872, 878, 887, 1571, 1604n, 1605n, 1720 Zeira, J., see Harrison, P. 875 Zejan, M., see Blomstrom, M. 627, 656, 659 Zeldes, S. 515 Zelizer, V.A. 1246n Zemsky, P., see Rob, R. 1659 Zervos, S., see Levine, R. 653, 655, 661, 886, 893, 894, 895n, 907, 907n Zeufack, A., see Bigsten, A. 1650 Zhang, J. 1602n Zhang, X. 1706n Zhao, R., see Parente, S.L. 1414 Zhu, X., see Restuccia, D. 720, 721, 723n Ziegler, D., see Pollard, S. 910n Zietz, J. 643
I-36 Zilibotti, F. 150n, 160n, 318n Zilibotti, F., see Acemoglu, D. 71n, 99, 102n, 120n, 130, 131, 136, 138n, 140, 142, 143, 150, 150n, 154n, 223n, 277n, 326, 357, 398, 428, 500n, 503, 541n, 607, 734n, 871, 871n, 876, 880, 883, 943n, 1000, 1425n, 1490n Zilibotti, F., see Aghion, P. 986n Zilibotti, F., see Doepke, M. 230, 260, 261, 1266 Zilibotti, F., see Marimon, R. 1345 Zillibotti, F., see Hassler, J. 1597n, 1617n Zilsel, E. 1138n Ziman, J. 1147, 1150n
Author Index Zingales, L. 883 Zingales, L., see Dyck, A. 917, 918n Zingales, L., see Guiso, L. 896, 896n, 908, 923, 1678, 1682, 1684 Zingales, L., see Kumar, K.B. 917n Zingales, L., see Rajan, R.G. 882, 885, 887, 910, 912, 912n, 923 Zinman, J., see Bertrand, M. 522 Zoabi, H., see Hazan, M. 231n, 259 Zoido-Lobaton, P., see Friedman, E. 999 Zoido-Lobaton, P., see Kaufmann, D. 649 Zou, H.-F., see Clarke, G. 921n Zou, H.-F., see Li, H. 462, 541, 655, 662
SUBJECT INDEX
1870 Education Act 209 1902 Balfour Act 209 β-convergence 582, 586, 828 σ -convergence 582, 828
Asian financial crisis 969 assembly lines 1310 assortative matching 1324, 1335 asymmetric information 981 Atlantic trade 217 autarky 1422, 1426, 1436, 1437, 1439, 1441, 1443–1446, 1452, 1456, 1463, 1464, 1466, 1472, 1485–1488 average labor productivity 769, 778, 800
abatement 1755, 1773 abatement intensity 1776 absorptive capacity 78 Académie Royale des Sciences 1137, 1143 access costs 1121 accumulation mechanism 1600 accumulation technology 1605 acquiring and processing information 871 acquis communautaire 1008 Adam Smith 1134 adaptation 1000 adaptive efficiency 1127 Adolf Wagner 1741 adoption and imitation R&D 857 adult mortality rate 708 advantage of backwardness 79 adverse selection 1720 age of the leadership 1208 agglomeration effects 326, 1473, 1481, 1483, 1484, 1486, 1487 aggregate demand 503 aggregate production function 475, 1383, 1391 aggregation 543 agriculture 229, 275, 443, 683, 721, 1227 air pollutants 1761 AK model 10, 823, 1017, 1104, 1779 Albrecht von Haller 1140 Alessandro Volta 1151 Alexander Chisholm 1159 alternating current 1195 appropriate incentives 973 appropriate institutions 100 appropriate technology 500, 734 appropriate technology and development 136 arable land 725 arbitrage equation 1080 Archibald Cochrane 1142 Arthur E. Kennelly 1153 Arthur Young 1138
baby boom 1231 baby bust 1231 balanced growth path 1077, 1404, 1775, 1776, 1794 balanced growth rate 824 Balassa–Samuelson effect 1002 banking 343, 867, 868, 878, 881–886, 893, 895, 907, 913, 916, 918–920 bargaining power 497 barriers to international trade 1507, 1508, 1512 barriers to technology adoption 717, 818, 827 barriers to trade 1507–1510, 1512, 1532, 1538 BEA–IBM constant quality price index 751 benefits to backwardness 837 Benjamin Franklin 1140 Benjamin Huntsman 1128 Benjamin Thompson 1141 Berlin Academy 1143 Bertrand competition 345 “best-practice” 981 bifurcation 323 big push 340, 503, 973, 1001 Birmingham Lunar Society 1142 birthweight 709 black market premium 1019 borders 1501, 1502, 1504–1507, 1510, 1511, 1513, 1517, 1518, 1533, 1535 borrowing constraints 82, 358 buffer stock of children 227 Caldwell hypothesis 1265 capital accumulation 233, 890, 895, 901, 1021, 1788 I-37
I-38 capital consumption 787 capital deepening 777 capital flight 994 capital flows 505, 533 capital income tax 845 capital market imperfections 872 capital markets 234, 477, 509 capital quality 761, 791 capital service flows 761 capital share xxiv, 686, 696, 1021, 1379, 1382, 1387, 1388, 1394, 1397, 1399, 1405 capital stock 791, 891, 1024, 1790 capital taxation 21 capital-embodied technical change 1292, 1333, 1346 capital-embodied technological change 1279 capital–output ratio 484 capital-skill complementarity 257, 1298, 1307, 1308, 1361 carbon monoxide 1772 carded wool and linen 1130 Carl Linnaeus 1135 Cavendish 1151 central bank independence 974 centralized bargaining 1326, 1329 CES 683 child labor 230, 373, 1228 child labor law 230 chlorine bleaching 1127 choice of plant technologies 1392 city formation 1585 city sizes 1566, 1568, 1583 civil liberties 1006 Claude Berthollet 1128, 1133, 1141 Clean Air Act 1769 Coase Theorem 422 Cobb–Douglas production function 504, 683, 1379, 1382, 1383, 1391 cognitive ability 1312 Colin MacLaurin 1139 collateral 350 collective action 391, 1663 collective bargaining 1325 college premium 1282, 1286 colonial powers 1054 colonialism 217 Combination Act 1168 commercial capitalism 1171 commitment problem 390 common law 506 common property 422
Subject Index communications technology 751 comparative advantage 857 competition 7, 423, 1409, 1412 competitive fringe 73 competitive search 1339 complementarity 162, 365, 1598 complementarity in innovation 144 complete convergence 595 composition of capital 682 composition of the capital stock 711 computers 747 conditional convergence 591, 1046 confined exponential diffusion 939 conflict of interest 390 constant elasticity formulation 1800 constant returns to scale 477, 527, 825, 1018, 1290, 1501, 1504 constant returns to scale technology 1379 consumption 514, 1018, 1186, 1355 contract enforcement 973 convergence 81, 308, 337, 477, 577, 583, 804, 899, 937, 941, 997, 1400, 1403, 1413, 1605 convergence clubs 77, 81, 256, 938, 943 convergence hypothesis 561 convex neoclassical model 307 convex technology 15, 314 coordination 328, 330, 1645 coordination failure 503, 1001, 1329, 1645 Copernicus 306 core–periphery models 1578 corporate governance 870, 872, 875, 882–884, 922, 974 corruption 364, 506, 883, 974 cost of living 1473, 1474 country size 842, 1501–1506, 1509, 1510, 1513, 1514, 1516, 1518–1522, 1524–1527, 1530, 1531, 1538 Crawshays 1159 creative destruction 1070, 1332 credit constraints 205, 360, 509, 1600 credit market 303, 351, 477, 509 credit market imperfections 1720 crime 1713 criteria pollutant 1785 Crompton 1159 cross-country growth regression xxiii, 576, 971 cross-country income distribution 330, 597, 887, 920, 921 cross-country regressions 506, 900, 919 cross-sectional risk 876
Subject Index crude birth rates 190, 191, 201 crude marriage rates 191 culture 397, 922, 923, 1115, 1268 currency board 994 d’Alambert 1137 Daniel Defoe 1127 Darwinian methodology 265 Darwinian survival strategies 231 David Hartley 1136 David Mushet 1130 Davies Gilbert 1139 de facto political power 391 de jure political power 391 debt contracts 871, 883 debt sustainability 973 decentralization 1505, 1535, 1539 default 350, 355 demand for human capital 205 demand spillovers 341 democracy 392, 1006, 1409 demographic patterns 177, 190, 1100, 1147, 1231, 1706 deposit insurance 907, 908 depreciation rate 690 desired consumption 1204, 1217 deskilling 1160 destruction of steam technology 1191 deterministic cycles 158 deunionization 1326, 1328 devaluation 990 development 990, 1017 development accounting 499, 681 development trap 940 dictatorship 392 Diderot 1136, 1137 diffusion 1136 diffusion lags 1194 diffusion of technology 1598 diminishing returns 308, 503, 523, 825, 1021, 1430–1436, 1439, 1442, 1444–1446, 1453, 1455, 1456, 1461, 1464, 1466, 1472, 1490 directed technical change 130, 1306, 1312, 1322 distance to the technological frontier 99 distribution of entries across sectors 1202 distribution of firm sizes 531 distribution of ratings for issues of new corporate bonds 1214 distribution of wealth 476 distributional conflict 1609
I-39 distributional dynamics 582 divergence 635 diversification 357, 870, 875, 876, 1001 diversified metro areas 1575 double their per capita income 1403 dual economy models 1577 dualism 1726 Dugald Stewart 1134 dynamic efficiency 986 dynastic construct 1380 earnings compression 1603 earnings volatility 1331 economic development 867, 1706 economic geography 1547 economic institutions 389 economic integration 1436, 1438–1441, 1453, 1490, 1511–1513, 1536 economic losers xxi, 434 economies of scale 1503, 1512, 1538 Edgeworth–Pareto substitutes 1252 education 102, 490, 496, 1116, 1155, 1706 education finance 1603, 1606 education premium 1300, 1303, 1336, 1355 education reforms 230 effective knowledge 1124 efficiency 531, 687, 1399, 1645 efficient institutions xxi, 422 elasticity of female labor supply 1267 elasticity of substitution 8, 728, 1027, 1379 electricity 1184, 1197, 1310 emission intensity of GDP 1818 emissions 1753, 1769 employment rate 1289, 1352 Encyclopedias 1137 Encyclopédie 1137 endogenous fluctuations 157 endogenous growth theory 558, 820, 1021, 1079, 1778 endogenous institutions 1623 endogenous skill-bias 132 endogenous supply of skills 1307 endpoints of a GPT era 1194 energy 1755, 1789, 1790 enforcement of contracts 1032 entrepreneurs 523, 998, 1119 entrepreneurship 998 entry, exit and mergers 1186, 1204 environment 1751
1450,
1238,
1354,
1297,
1017,
I-40 Environmental Catch-up Hypothesis 1811 environmental control costs 1761 Environmental Kuznets Curve (EKC) 1752 environmental quality 1751, 1758 epistemic base 1123 equity markets 878, 920, 1650 eras of GPT adoption 1208 Erasmus Darwin 1135, 1140 escape competition effect 88 estimates of aggregate relative efficiency 1394 ethnic differentials 1035 ethnic groups 1034 ethnic heterogeneity 1681 ethnolinguistic fractionalization 954, 1055 European colonialism 396 European mortality 417 Eurosclerosis 1612 event study approaches xxiii, 636 evolution 231 evolution of inequality 1604 evolution of the size distribution of cities 1549 evolutionary advantage 232 ex-post real interest rates 1220 exchange 881 exchangeability 581, 1663 exhaustible natural resource 1788 exhaustible resource 1755 exogenous technological progress 1774 expanding variety 116 expectational indeterminacy 162 expectational stability 163 experience premium 1285, 1313–1315, 1318, 1319 experimental philosophy 1163 exponential diffusion 939 export-processing zone (EPZ) 984 expropriation 394, 402 extended family system 1742 extended reproduction 1703 extent of the market 1514 external finance 910, 911, 914, 915 external financing constraints 868 externalities 326, 476, 502, 817, 819, 1000, 1020, 1503, 1504, 1644 extractive institutions 1054 extrinsic mortality risk 275 factor accumulation 317, 493, 1786 factor neutral 728 factor price equalization 1446, 1448, 1452, 1453, 1456, 1459, 1461, 1464–1466, 1472, 1476, 1480
Subject Index factor prices 212 factor shares 729 factor-biased innovation 131 factor-specific productivity 1289 factor-specificity 1360 factory system 306, 1160 family size 1737 farmers 1705 female education 234 female labor force participation 233, 1229 female–male earning ratios 1730 female’s age of marriage 191 fertility 174, 177, 182, 183, 191, 226, 373, 1072, 1104, 1115, 1148, 1227, 1233, 1377, 1378, 1381–1383, 1386, 1399, 1400, 1710, 1737, 1794 feudal 440 feudalism 440 finance companies 918 financial contracting 528 financial development 81, 150, 504, 890, 891, 898, 901, 911, 916, 919, 921, 1033, 1220 financial development indicators 890 financial functions 869, 870 financial instruments 869 financial intermediaries 871, 889, 890, 898, financial liberalization 974 financial markets 890, 918 “financial restraint” 981 financial sector reform 979 financial structure 920 first product or process innovation 1210 first-generation idea-based growth models 1092 fiscal decentralization 1564 fiscal policy 21, 923, 1033 fiscal system 1606 fix-price 1715 fixed costs 299, 340, 527 flexible technologies 1598, 1613, 1619 food-canning 1150 foreign exchange 1020 formal institutions 1005, 1671, 1713 fractionalization 1505 Francis Bacon 1133, 1134 Francis Home 1138 Francis Jeffrey 1134 Francis Upton 1153 free riding 873, 875, 882, 1661 free trade 1034 free trade club 1408, 1409, 1411–1413
Subject Index frictional inequality 1331, 1333 frictions in the labor market 1280 frontier technology 100 functional form 538 functional specialization 1575 functioning of the financial system 890 gas-lighting 1127 gender differences 1706 gender gap 232, 1266 gender wage gap 1285, 1312 General Electric (GE) 1211 General Motors (GM) 1211 general purpose technologies 1107, 1295, 1313, 1721 general skills 1337 genetic evolution 260 geography 277, 337, 339, 349, 397, 609, 620, 922, 923, 1054, 1554 George Campbell 1136 George Cayley 1140 George Melville 1153 George Stephenson 1141 Gibrat’s Law 1554 Gini coefficients 536, 920 glass industry 1127 globalization 280, 1426, 1436–1439, 1441– 1444, 1446, 1448, 1451, 1452, 1454– 1456, 1463, 1464, 1472–1475, 1480–1482, 1485–1488, 1490, 1491 Gompertz growth model 947 government failures 505, 998 GPT eras 1185 gradualism 989 Grande Encyclopédie 1136 Great Depression 1262 Great Divergence 174 Green Solow model 1754 growth accounting 314, 1101 growth disasters 322, 567 growth miracles xxiii, 322, 565, 1389, 1397, 1403, 1412 growth rates 819 growth slowdown xxiii, 567, 825 growth strategies 971 guilds 1162 Gustave Adolphe Hirn 1150 Gustave-Gaspard Coriolis 1159 Harberger triangles 1030 health 265, 490, 682, 1706
I-41 hedonic model 750, 751 Henry Cort 1139 Henry George Theorem 1567 Hermann Claudius 1153 heterodoxy 1004 heterogeneity 491, 1502, 1505, 1506, 1510– 1513, 1533, 1538, 1611, 1736 heteroskedasticity 640 “Hicks-neutral” 767 high-quality institutions 1005 high-school movement 1311 Hirschman–Herfindahl index 1556 historical self-reinforcement 317, 355 HIV/AIDS 971 home goods 1252 Hong Kong 984 horizontal innovation 114 horsepower in manufacturing 1187, 1188 hours worked 695 household appliances 1312 household capital 1252 household production 1227 household products 1258 housework 1251 human capital 174, 178, 340, 354, 475, 491, 682, 786, 806, 818, 872, 878, 938, 949, 1017, 1067, 1104, 1116, 1148, 1230, 1318, 1360, 1504, 1517, 1597, 1717 human capital externalities 502, 821, 1518 human capital promoting institutions 278, 280 humoral theory of disease 1124 Humphry Davy 1133, 1141, 1152 idea production function 1107 idea-based growth models 1090 ideas 819, 1065 identity 1654 imitation 73, 100, 127, 857 imitation costs 946 impatience traps 373 imperfect competition 340, 1082 import licenses 1031 import tariffs 1002 import-substituting industrialization (ISI) 984 import-substitution 973, 1034 improvement in the efficiency of the GPT 1195 incidental institutions xxi, 425 income disparity 304 income tax 1019 incomplete contracts 429, 518
I-42 increasing returns 299, 476, 501, 623, 1018, 1066, 1068, 1073, 1086, 1106, 1501, 1504, 1799 indivisibilities 357 indivisible projects 880 induced innovation 1798 Industrial Enlightenment 306, 1134 industrial policies 975 Industrial Revolution 176, 306, 877, 1098, 1115 industrialization 174, 1000, 1227, 1504, 1531 industry-level studies 919 inequality 212, 302, 536, 1597, 1600 infant and child mortality rates 225 infant mortality rates 226 inflation 903, 916, 1027 informal credit 480 informal sector 1026 information asymmetries 871, 878, 885, 922 information technology 746, 755, 756, 765, 771, 797, 798, 1184, 1194, 1279, 1293, 1301, 1303, 1320 informational complementarities 1321 initial conditions 320, 335, 598 initial public offerings (IPOs) 1200 innovation 27, 82, 100, 101, 857, 881, 1000, 1119, 1185, 1210 institutions 6, 7, 69, 277, 301, 364, 388, 506, 579, 867, 923, 973, 997, 1007, 1032, 1054, 1088, 1107, 1115, 1162, 1166, 1516, 1536, 1711 insurance 303, 351, 512, 878 insurance companies 918 intangible capital 1393, 1394 integrated circuits 747 integrated economy 1425, 1426, 1430, 1431, 1433, 1437–1439, 1441, 1443, 1444, 1446–1450, 1453, 1455–1457, 1465, 1472, 1475, 1476, 1482, 1483, 1488 intellectual property rights 128, 1119, 1165 inter-temporal optimization 1719 interest rates 479, 1204 intermediaries 869, 890, 918 internal combustion 1196 internal finance 910 international diffusion of knowledge 821 international externalities 818 international financial codes and standards 974 International Monetary Fund 298 international trade 216, 281, 1425, 1446, 1472, 1473, 1490
Subject Index internationally harmonized prices 785, 806 Internet 751 interstate commerce clause 1408 intertemporal elasticity of substitution 1032 intertemporal externality 117 intrinsic mortality risk 275 inventions 1119 investment xxvii, 101, 197, 478, 493, 506, 819, 914, 998, 1018, 1106, 1200, 1397, 1400, 1407 Jacquard loom 1127 Jacques-François Demachy 1137 James Keir 1128, 1136, 1151, 1159 James P. Joule 1139, 1149 James Watt 1136, 1159 Japanese “model” 1007 Jean-Antoine Chaptal 1128 Jean-Victor Poncelet 1159 Jesse Ramsden 1128 job destruction 1332, 1335, 1346 John Coakley Lettsom 1140 John Ericsson 1140 John Farey 1127 John Harrison 1128 John Herschel 1135 John Kay 1159 John Playfair 1163 John Roebuck 1159 John Smeaton 1128, 1157, 1159 John Whitehurst 1137, 1159 joint stock company 879 Jonas W. Aylsworth 1153 Joseph Banks 1140 Joseph Black 1128, 1136, 1139 Joseph J. Lister 1152 Joseph Priestley 1133, 1136 Josiah Wedgwood 1151 just-in-time production 1409 K and r strategies 231 kinship xx, 364, 367 knowledge 1017, 1071 knowledge externalities 87, 818, 1105, 1566, 1605, 1799 knowledge or technology component 1390 Kuznets hypothesis 1282 Kyoto Protocol 1792 labor augmenting technological progress 1773 labor force participation 233, 1285, 1706 labor inputs 761, 776, 805
Subject Index labor market 1340, 1353, 1606, 1712 labor mobility 1342 labor productivity 527, 1410, 1411 labor quality 763, 794 labor share 1288, 1347 labor-augmenting technological change 1022 Laggards 1402 Lagrange stability 377 laissez-faire 120, 994, 1597, 1601 land productivity 183 land–labor ratio 183 language 1121 late starters 1376, 1396, 1401 law and finance 887 law and order 1117 leadership 1648, 1657 learning 516, 1360 learning and sunspots 162 learning externalities 821, 999 learning-by-doing 821, 1000, 1295, 1314, 1334 Leblanc soda making 1127 legal system 508, 868, 869, 886, 887, 898, 919, 922, 923 less developed economies 174 life expectancy 185, 276, 954, 1704 life insurance 897 lifetime earnings 1354 limited enforcement of contracts 1355 limited patent protection 122 limits to growth 1751, 1788 liquidity 355, 872, 875–877, 882, 894, 896, 897 literacy rates 197 loan renegotiations 884 lobbying 1629 local financial development 896, 908 local government 1688 local markets 1473, 1487, 1489 localization economies 1565 logistic diffusion 939 logistic technology diffusion 942, 946 Lombe brothers 1149 Louis Navier 1159 Louis Pasteur 1150 Louis-Jacques Goussier 1137 love of variety 822 Luddite rebellion 1167 Ludwig Prandtl 1153 machine-breaking 1167 macroeconomic policies 986, 1017, 1033 macroeconomic volatility 879, 1045
I-43 macroinventions 1129 malnutrition 373 Malthusian models 174, 180, 182, 252, 256, 855, 1099, 1378, 1781 Manchester Literary and Philosophical Society 1142 manufacturing 911, 1229 Marc I. Brunel 1159 marginal abatement costs 1786 maritime technology 1118 market capitalization 895, 916 market frictions 349, 867, 869, 873, 999 market integration 1739 market liberalization 993 market power 343, 345 market size 1430, 1432–1436, 1438–1440, 1444–1446, 1450, 1452, 1453, 1455, 1456, 1459, 1461, 1464, 1472, 1487–1490, 1501, 1504, 1505, 1514, 1516, 1518, 1519, 1532 marketization 1739 markets 868, 869, 881, 883, 885, 886, 890, 893, 918, 919, 973, 994, 1119 marriage age 191 marriage institutions 231 Marshallian dynamics 328, 366 mass production 1311 master–apprentice relationship 1128 matching 340, 346, 1332, 1338, 1346, 1347, 1361 Matthew Boulton 1162 mechanization 1229 megacities 1550, 1583 memory chips 748 mercantilism 1126, 1162 metallurgical knowledge 1132 Michael Faraday 1156 microinventions 1125 microprocessors 748 microscope 1152 Microsoft Corporation 750, 1212 migration 369, 1712 Mincerian returns to education 484, 700, 1727 minimum wage 1603 misallocation of capital 478 model uncertainty 559, 612, 613 Modern Growth Regime 176 Modified Political Coase Theorem 424 monitoring 870, 875, 1742 moral hazard 981, 1322, 1720 mortality 184 multi-sector growth models 4, 5
I-44 multifactor productivity 784 multiple regimes 145, 150, 162, 335, 621, 622, 727, 1021, 1338, 1659 natural experiment 396 natural resources 1751, 1791 natural selection 231 Nelson–Phelps hypothesis 937, 1301 neoclassical models xx, 15, 307, 475, 494, 853, 1021, 1022, 1374 Neolithic period 275 network externalities 339, 349, 368, 1189 network formation 1659 Nevil Maskelyne 1158 new good externalities 819 Nicolas Appert 1150 nitrogen oxide 1772 non-aggregative growth models 478, 505, 535 nonconvex growth 319, 353 nonrivalry 857, 1065, 1072, 1086 nonrivalry of ideas 1073, 1106 nontraded goods 1473, 1519, 1520 norms 302, 1644 North American Free Trade Agreement 1408 North–North trade 216 North–South trade 216 occupation-specific human-capital 1336 old age insurance 369 old-age security hypothesis 234 on-the-job training 1318 open knowledge 1121 open science 1132 openness to international trade 1032 opportunity cost of capital 510 opportunity cost of children 228 optimal allocation 1080 optimal scale 531, 1310 organization of education 102 organizational capital 1206, 1296, 1325 organizational structure 1321, 1598, 1613 outsourcing 1574 outward-orientation 973 own-account software 754 parental education 700 partial convergence 584 patents 69, 1082, 1105, 1165, 1166, 1186, 1198 Penn World Tables 685, 804 Philippe LeBon 1128 physical capital 682, 818, 895, 901, 1019, 1393, 1394
Subject Index Pierre Louis Guinand 1127 Pierre-Simon de Laplace 1151 pivotal voter 1608 policy 69, 355, 868, 886, 1017, 1054, 1093, 1769 political factors xxi, 214, 389, 390, 423, 432, 868, 922, 923, 1599, 1600, 1606, 1609 politico-economic steady states 1599 pollution 1755, 1766, 1816 pollution abatement 1763, 1779 Post-Malthusian Regime 177, 186 poverty 537, 622, 920, 921, 986, 1017, 1021, 1146 pre-demographic transition era 235 precautionary demand for children 227 predation 394 preference for offspring’s quality 231 prescriptive knowledge 1122 price liberalization 979 price taking 314 primary electric motors 1187 printing press 1121 prisoner’s dilemma 1654 private good 1018 private property 422 private rate of return 845 product cycle 1576 product cycle trade 127 product-market competition 69, 718 production function 528, 768, 1280 productivity 87, 136, 217, 316, 475, 491, 500, 527, 747, 783, 890, 895, 901, 1184, 1186, 1203, 1204, 1227, 1295, 1346 property rights 15, 388, 508, 914, 973, 1032, 1818 propositional knowledge 1121, 1123 protectionism 1515, 1532–1534 prudential regulation 974 public education 194 public enterprises 975, 1001 public goods 1503, 1505, 1510–1512, 1531, 1538, 1648, 1649, 1703 quality of goods 67, 69, 714, 782, 791, 822, 1292 quantity–quality trade-off 231, 1239 R.J. Petri 1152 Ramsey–Cass–Koopmans model 587 Realschulen 1164 redistributive institutions 1597, 1614
Subject Index redistributive policies 1359 Reform Act of 1884 215 Reform Acts of 1867 and 1884 214 regulations 508, 868, 869, 886, 922, 923, 1787, 1816 relative efficiency 1399 relative price of equipment 1196, 1298, 1308 relative prices 1027, 1713 relative wages of women 225 religion 194, 400, 923 René Réaumur 1124, 1139 rentiers 1705 rents 394 replacement level 191 replication argument 1066 reputation 367, 518 research and development 70, 120, 712, 818, 833, 841, 845, 854, 879 residual inequality 1361 Richard Roberts 1128, 1159 Richard Trevithick 1139, 1159 rise in the factory 1169 risk 157, 227, 355, 368, 513, 870, 876, 922 Robert continuous paper-making 1127 Royal Society 1135 Sadi Carnot 1149 savings 520, 867, 901, 1380, 1704, 1719 scale effects 92, 124, 501, 837, 999, 1088, 1089, 1095, 1096, 1101, 1516–1518, 1530, 1531, 1538, 1752, 1757 schooling 197, 496, 682, 828, 954, 1227, 1713 Schumpeterian effect of product market competition 88 Schumpeterian growth paradigm 69 scientific management 1230 scientific revolution 270, 1131 secession 1506, 1513, 1514, 1536, 1538 Second Industrial Revolution 178, 225, 1229 second-generation reforms 974 sector effects 47, 682, 1289, 1291, 1703, 1713, 1739 semi-endogenous growth theory 96 semiconductors 746, 747, 1291 shock therapy 989 Simon Newcomb 1153 size distribution of cities 1548 size of countries 1502, 1505, 1510, 1512, 1514, 1530 size of the public sector 1703
I-45 skills
134, 205, 229, 239, 257, 281, 1204, 1205, 1227, 1240, 1279, 1283, 1297, 1298, 1300, 1303, 1306–1308, 1310, 1311, 1314, 1315, 1321, 1327, 1329, 1334, 1338, 1344, 1349, 1354, 1361, 1597, 1598, 1613, 1616, 1620, 1703, 1705 Smithian growth 1116 social capital 402, 923, 1020, 1641 social contract 1597, 1609 social differentiation 1705 social exchange 1645 social institutions 1705 social insurance 1006, 1600, 1741 social interactions 1673 social learning 516 social mobility 1605 social networks 1644 social policy 986 social preferences 1007 social rate of return 838, 1087 social relations 1703 social safety nets 974 social structure 1659 socialist economic system 1050 Society of Arts 1142 sociopolitical stability 194 Solow model 17, 580 Solow regime 159 Solow residual 494, 681 Solow variables 580 South Asia 307, 971 specialization 299, 349, 823, 881, 1504 spillovers 327 static efficiency 986 Statute of Artificers 1168 steady states 578, 1018 steampower 1129, 1130, 1191 steelmaking 1130, 1132 stock markets 874, 877, 882, 886, 893–895, 912, 1206 stock of capital 198 stock of knowledge 1796 stock prices 1204 Stone–Geary preferences 1032 structural evolution 976, 1263, 1352, 1451, 1452, 1455, 1456, 1461, 1463, 1464, 1466, 1470, 1480 stylized facts 561, 1771 Sub-Saharan Africa 175, 304, 307, 971, 1703 subsistence level 179 sustainability 174, 997, 998, 1753, 1774
I-46
Subject Index
symmetry-breaking 362 systems of cities 1547, 1565
trust 1117, 1643, 1644, 1646 twin-peaks 596
targeted anti-poverty programs 974 taxes 869, 979, 1018, 1019, 1380, 1603 Tayloristic organization 1311, 1321, 1324, 1330 teacher–pupil ratios 698 technical externalities 300 technical progress 1705, 1720 technique 1120, 1752, 1758 technische Hochschulen 1164 technology 77, 83, 140, 150, 174, 178, 190, 239, 339, 373, 478, 496, 499, 501, 516, 577, 818, 867, 871, 876, 937, 941, 1017, 1022, 1116, 1131, 1144, 1204, 1217, 1227, 1229, 1256, 1304, 1308, 1337, 1339, 1350, 1392, 1598, 1627, 1628, 1754, 1756, 1775 terms of trade 835 The Reform Act of 1832 214 theory of aggregate production function 1391 theory of relative efficiencies 1389 theory of TFP 1375, 1390 theory open-endedness 639 thermodynamics 1149 Thomas Alva Edison 1153, 1184, 1250 Thomas Jefferson 1069, 1137 throstles 1129 tightness 1125, 1152 Tobern Bergman 1124 Tobias Mayer 1158 total factor productivity 478, 499, 605, 607, 684, 767, 835, 940, 1289, 1291, 1375, 1379, 1380, 1382 trade 842, 974, 1001, 1050, 1106, 1186, 1204, 1218, 1307, 1504–1506, 1510, 1513, 1514, 1516, 1518, 1519, 1530, 1538, 1817 transitions 176, 332, 494, 980, 1026, 1073, 1374, 1375, 1549, 1775 Treaty of Rome 1410
unemployment 696, 1279, 1288, 1328, 1331, 1340, 1345, 1347 unskilled labor 1229, 1705 urban growth 193, 408, 1546, 1554, 1560, 1561, 1581, 1706, 1712 vintage capital 1300, 1315, 1332, 1334, 1336, 1337, 1347, 1361 voluntary organizations 1648 wages 131, 142, 213, 232, 262, 686, 1158, 1279, 1281, 1327, 1328, 1340–1342, 1347, 1352, 1629 Wagner’s law 1706 Washington Consensus 973, 1033 welfare states 1597, 1601 Western Offshoots 175, 186 Wilbur Wright 1153 William Cruickshank 1151 William Cullen 1138 William Ellis 1137 William Murdoch 1159 William Nicholson 1140, 1152 William Petty 1069 William Rankine 1139, 1150 William Thomson 1150 wind power 1129 women’s liberation 1250 World Bank 1730 world technology frontier 836 World War II 1262, 1725 worldwide changes in patent policy 1199 Wright brothers 1153 wrought iron 1129 Zipf’s Law 1548
